[9 C( O. Measuring Economic Benefits for Water Investments and Policies. k-p 33. RobertA. Young. Public Disclosure Authorized

Transcription

1 Work in progress for public discussion WORLD BANK TECHNICAL PAPER NO. 338 k-p 33 [9 C( O Measuring Economic Benefits for Water Investments and Policies Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized RobertA. Young

2 RECENT WORLD BANK TECHNICAL PAPERS No. 263 Le Moigne, Subramanian, Xie, and Giltner, editors, A Guide to the Formulation of Water Resources Strategy No. 264 Miller and Jones, Organic and Compost-Based Growing Media for Tree Seedling Nurseries No. 265 Viswanath, Building Partnerships for Poverty Reduction: The Participatory Project Planning Approach of the Women's Enterprise Management Training Outreach Program (WEMTOP) No. 266 Hill and Bender, Developing the Regulatory Environment for Competitive Agricultural Markets No. 267 Valdes and Schaeffer, Surveillance of Agricultural Prices and Trade: A Handbookfor the Dominican Republic No. 268 Valdes and Schaeffer, Surveillance of Agricultural Prices and Trade: A Handbookfor Colombia No. 269 Scheierling, Overcoming Agricultural Pollution of Water: The Challenge of Integrating Agricultural and Environmental Policies in the European Union No. 270 Banerjee, Rehabilitation of Degraded Forests in Asia No. 271 Ahmed, Technological Development and Pollution Abatement: A Study of How Enterprises Are Finding Alternatives to Chlorofluorocarbons No. 272 Greaney and Kellaghan, Equity Issues in Public Examinations in Developing Countries No. 273 Grimshaw and Helfer, editors, Vetiver Grass for Soil and Water Conservation, Land Rehabilitation, and Embankment Stabilization: A Collection of Papers and Newsletters Compiled by the Vetiver Network No. 274 Govindaraj, Murray, and Chellaraj, Health Expenditures in Latin America No. 275 Heggie, Management and Financing of Roads: An Agendafor Reform No. 276 Johnson, Quality Review Schemesfor Auditors: Their Potentialfor Sub-Saharan Africa No. 277 Convery, Applying Environmental Economics in Africa No. 278 Wijetilleke and Karunaratne, Air Quality Management: Considerationsfor Developing Countries No. 279 Anderson and Ahmed, The Casefor Solar Energy Investments No. 280 Rowat, Malik, and Dakolias, Judicial Reform in Latin America and the Caribbean: Proceedings of a World Bank Conference No. 281 Shen and Contreras-Herrnosilla, Environmental and Economic Issues in Forestry: Selected Case Studies in Asia No. 282 Kim and Benton, Cost-Benefit Analysis of the Onchocerciasis Control Program (OCP) No. 283 Jacobsen, Scobie and Duncan, Statutory Intervention in Agricultural Marketing: A New Zealand Perspective No. 284 Valdes and Schaeffer in collaboration with Roldos and Chiara, Surveillance of Agricultural Price and Trade Policies: A Handbookfor Uruguay No. 285 Brehrn and Castro, The Marketfor Water Rights in Chile: Major Issues No. 286 Tavoulareas and Charpentier, Clean Coal Technologiesfor Developing Countries No. 287 Gillharn, Bell, Arin, Matthews, Rumeur, and Heam, Cotton Production Prospectsfor the Next Decade No. 288 Biggs, Shaw, and Srivastiva, Technological Capabilities and Learning in African Enterprises No. 289 Dinar, Seidl, Olem, Jorden, Duda, and Johnson, Restoring and Protecting the World's Lakes and Reservoirs No. 290 Weijenberg, Dagg, Kampen Kalunda, Mailu, Ketema, Navarro, and Abdi Noor, Strengthening National Agricultual Research Systems in Eastern and Central Africa: A Frameworkfor Action No. 291 Valdes and Schaeffer in collaboration with Errazuriz and Francisco, Surveillance of Agricultural Price and Trade Policies: A Handbook for Chile No. 292 Gorriz, Subramanian, and Simas, Irrigation Management Transfer in Mexico: Process and Progress No. 293 Preker and Feachem, Market Mechanisms and the Health Sector in Central and Eastern Europe No. 294 Valdes and Schaeffer in collaboration with Sturzenegger and Bebczuk, Surveillance of Agricultural Price and Trade Policies: A Handbookfor Argentina No. 295 Pohl, Jedrzejczak, and Anderson, Creating Capital Markets in Central and Eastern Europe No. 296 Stassen, Small-Scale Biomass Gasifiersfor Heat and Power: A Global Review No. 297 Bulatao, Key Indicatorsfor Family Planning Projects No. 298 Odaga and Heneveld, Girls and Schools in Sub-Saharan Africa: From Analysis to Action (List continues on the inside back cover)

3 WORLO BANK TECHNICAL PAPER NO. 338 Measuring Economic Benefits for Water Investments and Policies RobertA. Young The World Bank Washington, D. C.

4 Copyright i 1996 The International Bank for Reconstruction and Development/THE WORLD BANK 1818 H Street, N.W. Washington, D.C , U.S.A. All rights reserved Manufactured in the United States of America First printing September 1996 Technical Papers are published to communicate the results of the Bank's work to the development community with the least possible delay. The typescript of this paper therefore has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibility for errors. Some sources cited in this paper may be informal documents that are not readily available. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility whatsoever for any consequence of their use. The boundaries, colors, denominations, and other information shown on any map in this volume do not imply on the part of the World Bank Group any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to copy portions for classroom use is granted through the Copyright Clearance Center, Inc., Suite 910, 222 Rosewood Drive, Danvers, Massachusetts 01923, U.S.A. The complete backlist of publications from the World Bank is shown in the annual Index of Publications, which contains an alphabetical title list (with full ordering information) and indexes of subjects, authors, and countries and regions. The latest edition is available free of charge from the Distribution Unit, Office of the Publisher, The World Bank, 1818 H Street, N.W., Washington, D.C , U.S.A., or from Publications, The World Bank, 66, avenue d[iena, Paris, France. Robert A. Young is Professor Emeritus at Colorado State University. ISBN ISSN: Library of Congress Cataloging-in-Publication Data Young, Robert A. (Robert Alton), Measuring the economic benefits for water investments and policies / Robert A. Young. p. cm. - (World Bank technical paper, ISSN ; no. 338) Includes bibliographical references. ISBN Water resources development-economic aspects-mathematical models. I. Title. II. Series. HD1691.Y '001'51-dc2O CIP

5 TABLE OF CONTENTS FOREWORD ACKNOWLEDGMENTS EXECUTIVE SUMMARY Background, Objectives and Scope Concepts and Methods Methods of Nonmarket Valuation of Water Methods of Valuing Intermediate Goods Valuing Water as a Private Consumption Good Valuing Public Good Benefits of Water Conclusion GLOSSARY vii viii ix ix x x x xi xii xiii xiv 1. INTRODUCTION The Need for Synthetic Estimates of the Economic Benefits of Water-Related Decisions The distinctive nature of water supply and demand Objective, Scope and Plan of the Monograph 6 2. CONCEPTUAL FRAMEWORK AND SPECIAL PROBLEMS IN VALUING WATER Introductory Comments Economic Value versus Other Concepts of Value Economic Criteria for Resource Allocation and Valuation The Pareto Principle and Economic Efficiency From Theory to Practice Economic Valuation in the Absence of Market Prices The Need for Shadow (Accounting) Prices Defining Shadow Prices: The Willingness to Pay Principle Economic Surplus and Measures of Benefit Opportunity Costs: Measuring Foregone Benefits of Reduced Water Use 17 iii

6 2.5 Some Specific Cases of Economic Evaluation of Water Resource Issues Evaluating Investments in Additional Water Supplies Evaluating Proposals to Reallocate Water Among Sectors The Benefits of Improved Water Quality Benefits of Improved Water-Supply Reliability Other Conceptual Issues in Water Valuation The With-Without Principle Accounting Stance Special Problems Associated with Nonmarket Valuation of Water Long-mn versus Short-run Values The Value of Water Relative to Other Commodities Commensurability of Place, Form and Time Appropriate Allocation Variable: Depletion versus Withdrawal Uncertainty and Sensitivity Analysis TECHNIQUES FOR MEASURING THE VALUE OF WATER Preliminaries Observations of Transactions in Water and Related Goods Values from Rentals and Sales of Water Rights Valuing Water as a Part of a Bundle of Marketed Characteristics: The Hedonic Method Demand Functions from Water Utility Sales Data Techniques for Valuing Water as an Intermediate Good The Residual Imputation Approach Derivation of the Residual Value The "Change in Net Income" and other Variants of the Residual Approach General use of Residual Imputation Conceptual Problems Arising with Residual Imputation: Specifying the Production Function Assigning Prices to Inputs and Outputs Residual Imputation in Long-run versus Short-run Planning Contexts Mathematical Programming (Optimization) Models as an Application of the Residual Imputation Technique Erroneous Water Valuation with a Version of the Residual Technique: The "Value Added" Approach from Input-Output Models The Input-Output (Interindustry) Model Applied to Water Valuation Interpreting the "Value of Water" Derived from Value Added The "Alternative Cost" Approach Measuring Willingness to Pay For Water As a Consumption Good: Overview User Surveys to Determine Willingness to Pay for Water-Related Public Benefits Revealed Preference or Observed Indirect Methods 42 iv

7 3.7.1 The Travel Cost Approach The Hedonic Price Approach Questioning Consumers Regarding Valuation of Hypothetical Environmental Changes-The Contingent Valuation Approach Methodological Issues in Designing a CVM Study Forms of Questions Used in CVM Studies Potential Sources of Systematic Error (Bias) in Contingent Valuation Studies Advantages and Disadvantages of the Contingent Valuation Method The "Unit Day Value" Table: A Valid Shortcut Method for Measuring Recreational Benefits? More Formal Approaches to Generalized Values: Benefit Transfer and Meta-Analysis APPLICATIONS 1: THE CASE OF WATER USED IN INTERMEDIATE GOODS Measuring the Value of Water for Irrigated Crop Production Is Irrigation Water Accurately Valued? Applied Irrigation Water Valuation: The Residual Imputation Approach and Variations The "Change in Net Income" Method Specific Issues Arising in Application of Residual Imputation to Valuing Irrigation Water Implementation of the "With versus Without" Test in the Residual and Change in Net Income Methods Mathematical Programming Models for Irrigation Water Valuation The Hedonic (Land Value) Approach Applied to Crop Irrigation The "Alternative Cost" Approach to Irrigation Water Value Measuring Benefits of Improved Quality of Irrigation Water Valuing Water in Industrial Uses The Econometric Approach to Industrial Water Demand Analysis An Approximation Using An Assumed Demand Elasticity Finding the Area Under the Demand Function Value of Raw Water per Volumetric Unit Adjustment for Water Delivery Losses Residual and Mathematical Optimization Models of Industrial Water Demand Change in Net Income Mathematical Programming Models for Estimating Industrial Water Value The "Value-Added" Approach Risks Overstating the Economic Value of Industrial Water Valuing Water in Hydropower Generation Introduction Production of Hydroelectric Power Distinctive Issues in Hydropower Evaluation A Model for Deriving the Value of Water in Hydropower Generation 82 APPENDIX TO CHAPTER A. 1 Introduction 85 v

8 4A.2 Shadow Pricing as a Component of Water Valuation 85 4A.3 Choosing the Social Rate of Discount 85 4A.4 The Shadow Wage Rate 87 4A.5 Calculating an Equivalent Uniform Annual Cost Or Benefit APPLICATIONS 2: VALUING WATER AS A PRIVATE CONSUMERS' GOOD Overview Valuing Water in Municipal Uses in Developed Countries Econometric Methods for Domestic Water Demand Estimation Type and Number of Observations Specifying the "Price" Variable: Marginal or Average Price? Additional Considerations in Specifying the Domestic Water Demand Function Other Methods of Estimating Residential Water Demand Deriving a Value of Raw Water for Domestic Purposes from a Demand Function Measuring Benefits of Domestic Water Supply Reliability Rural Domestic Water Demand in Developing Countries APPLICATIONS 3: VALUATION OF SOME INSTREAM PUBLIC GOODS: INSTREAM FLOWS, WATER QUALITY IMPROVEMENT AND FLOOD RISK REDUCTION Deriving the Value of Water in Recreational and Amenity Uses: Overview Valuing Instream Flows for Outdoor Recreation Valuation of Water Quality Improvement Recreational Benefits of Water Pollution Control Economic Benefits of Reducing Waterborne Health Risks the Economic Benefits of Allocating Water for Waste Dilution: an Alternative Cost Approach Benefits of Flood Risk Reduction 103 REFERENCES 104 vi

9 FOREWORD The 1993 World Bank Water Resources Management Policy has at its core the adoption of a comprehensive analytical framework and the handling of water as an economic good. The policy emphasizes the interrelationships between competing sectors and the importance of considering the economic value of water in various uses in water development and allocation decisions. Economic evaluation of water-related investments in development and management of irrigation, hydropower, urban and rural water supply and sanitation, and in flood control is important because it aids in determining the worth to people of proposed projects and in estimating the degree to which they are willing to pay for benefits. In the prevalent situation of constrained public budgets, conceptually correct and empirically valid estimates of the economic contribution of water in each water using sector are essential for making economically sound investment decisions. The literature has gained in recent decades a number of techniques for measuring the economic values or benefits associated with various uses of water via both market and nonmarket allocations. Progress with methods for estimating economic benefits in actual cases is also well advanced. This monograph reviews, assesses and describes the operational use of the concepts and methods for estimating economic benefits of investment and allocation decisions involving water. It is intended for project task managers who face the task of evaluating water-related allocation decisions. The aim is to provide such analysts with a practical, yet theoretically sound procedural handbook to help guide their analyses. Alexander F. McCalla, Director Agricultural and Natural Resources Department Anthony J. Pellegrini, Director Transportation, Water & Urban Development Department vii

10 ACKNOWLEDGMENTS Robert A. Young is Professor Emeritus, Colorado State University, Fort Collins, Colorado, 80523, USA. This monograph was supported in part by a consultancy contract with the Agricultural and Natural Resources Department of the World Bank. The author is particularly grateful to Lee Gray, whose earlier collaboration provided a basis for this effort. Thanks also to Ariel Dinar and Rita Cestti for their guidance and sharing of ideas during the completion of this monograph. K. W. Easter, David Hanrahan, Richard Reidinger and Dale Whittington provided helpful suggestions on earlier drafts. The monograph typesetting was prepared by Patricia Noel. Vill

11 EXECUTIVE SUMMARY BACKGROUND, OBJECTIVES AND SCOPE Water resources provide important benefits to humankind, both commodity benefits and environmental values. For physical, social and economic reasons, water is a classic nonmarketed resource Even for commodity uses, market prices for water are seldom available or when observable, often are subject to biases. Water-based environmental values and external effects are rarely, if ever, priced. However, because of the increasing scarcity of water for both its commodity and environmental benefits and scarcity of the resources required to develop water, economic evaluation plays an increasingly important role in public decisions on water projects, reallocation proposals and other water policies. As an alternative to market prices, shadow or accounting prices reflecting economic benefits or values are needed to provide a basis for water-related investment and allocation decisions. Economists have in recent decades developed and refined a number of techniques for measuring the economic values associated with nonmarket allocation in the subject matter areas relating to the environment and natural resources. These techniques call for a careful wedding of economic theory and applied economic practices. Implementation of any of these methods may require use of one or more elements from the applied economists' toolkit, including primary and secondary data collection, econometrics, discounted cash flow analysis and optimization models. This monograph is designed to provide a review, exposition and critical assessment of the concepts and methods for estimating economic benefits of investment and policy decisions involving water It is intended for field practitioners possessing some training in applied economics who face the task of evaluating water-related decisions. The aim is to provide such analysts with a practical, yet theoretically sound procedural handbook to guide their evaluations. The main focus of the report is on valuation of changes in water supply, although attention is also given to measuring benefits of increased reliability of water supply and to improved water quality. The report deals only with real effects, those reflecting positive or negative changes in economic welfare via actual changes in quantities of goods and services available, or changes in the amount of resources used. So-called secondary economic impacts, those reflected in changes in incomes or prices, (such as effected by increased purchases of goods and services in a regional economy), are not considered. The techniques of valuation often differ between intermediate goods or producers' goods (those employed to make products to be eventually used by consumers) and consumers' goods (those providing human satisfactions). The importance of this distinction is that the economic theory of a profit-maximizing producer provides the conceptual framework for the valuation of intermediate goods, while the theory of the individual consumer is the basis for valuing consumer goods. A further useful distinction is between private and public goods or services. Private goods are distinguished by the fact of being rival in use, meaning that one person's use of a unit of water necessarily precludes use by others of ix

12 that unit. The opposite end of the continuum is occupied by public goods; those that are nonrival in consumption, meaning that one person's use does not preclude enjoyment by others. The largest offstream uses of water by humankind are in intermediate good uses, mainly for crop irrigation and industrial uses. Residential water is an example of a final consumption good from the private (rival) good classification, while recreation and amenity services, including improved water quality, provide public good benefits. The main body of the monograph is organized into three parts. The initial chapters introduce the subject and present a sketch of the conceptual framework for nonmarket valuation of water. Next is a chapter which introduces the assumptions and procedures for implementing the various methods which have been developed for estimating waterrelated economic benefits. The remaining three chapters discuss the application of the various valuation methods to, respectively, intermediate goods, private consumer goods and public goods. This Executive Summary is limited to discussion of concepts and a brief review of the methods employed for valuing water. CONCEPTS AND METHODS Numerous concepts of value have been identified by philosophers. The economic concept of value is narrowly defined, referring to measures in money terms of welfare or satisfaction of human preferences. A person's welfare change from some proposed improvement is measured as the maximum amount of money a person would be willing to forgo to obtain the improvement. Conversely, for a change which reduces welfare, the measure is the amount of compensation required to accept the change. "Economic benefit" and "economic value" are used interchangeably in this monograph to refer to money measures of welfare changes resulting from investment projects or policy initiatives. (However, not all money measures are appropriate. Only measures of changes in economic surplus qualify as indicators of economic value). An individual's valuation of increased goods or services is posited to be a decreasing function of the amount of goods or services available. Valuations are also assumed to be influenced by other factors, including the costs of substitutes, incomes, and so on. The conventional demand or marginal benefit function is the concept measured in economic valuation exercises. Methods of Nonmarket Valuation of Water Among the several methods developed to estimate the value of water, some are best- suited for intermediate good cases, others for private consumer goods, yet others for public goods. Some apply to more than one case. Methods of Valuing Intermediate Goods Residual approaches.-for intermediate good uses of water-primarily in crop irrigation and industrial activities-models of the profit-maximizing firm are employed. The x

13 simplest of firm models employ a budgeting approach (often facilitated by spreadsheet software). The imputation process involves subtracting forecasted costs of nonwater inputs from forecasted revenues. The remaining or residual surplus of revenues over nonwater costs is imputed as the benefit or value of water to the firm. For cases where government policies distort market prices of inputs (e.g. labor or electricity) or outputs (e.g. subsidized agricultural crops), an adjusted "social" value usually should be calculated. More advanced analytic techniques-either with spreadsheets or mathematical programming-permit the calculation of the change in net income associated with changing water supply while incorporating refinements and extensions to obtain more empirical realism. Such extensions might include representation of how additional products, alternative levels of nonwater production inputs or of alternative water use technologies might affect the estimated benefits. The residual technique is subject to error; for example, the costs of certain nonwater inputs might be understated, or even omitted, thereby overestimating the residual value attributed to water. Mistakes of this sort have been found in regional input-output models, where the entire value-added (payments to wages, salaries, profits, interest, depreciation, other natural resources, etc.) has been incorrectly attributed to water. The alternative cost technique.-another method appropriate to estimating waterrelated intermediate good benefits (and also at times to consumption goods benefits) is the alternative cost approach. The technique can be applied under the assumption (valid only in certain limited instances) that if a given project of specified output costs less than the next- best public or private project which can achieve the same output, then the cost of the next best project can be assigned as the benefit to the public project under consideration. The alternative cost approach has been employed for evaluation of many types of waterrelated benefits, particularly hydroelectric power, (but also including domestic water use, thermal electric power and other industrial uses, and waste load dilution). When estimates of a direct demand schedule proves difficult because of lack of data or other reasons, the alternative cost approach may provide a solution. However, the analysis must verify that the higher cost alternative would itself actually be economically feasible in the absence of the project under consideration. Valuing Water as a Private Consumption Good Inferences from observed transactions in water.-for cases where actual observations on quantities consumed for varying prices of water, together with the corresponding data on other explanatory variables are available, an econometric approach is possible. This is not a true market example, because the water is typically sold by monopoly suppliers at a fixed price schedule. An abstract demand function is formulated which hypothesizes a connection between water consumption (the variable to be explained) and price and other factors influencing the dependent variable. Parameters of demand equations in this approach are estimated via statistical inference, usually multiple regression techniques. The main application is to municipal water demand. However, real prices in a single municipality seldom change enough to trace out a demand function, but data from a crossxi

14 section of water suppliers may exhibit sufficient variation. The econometric approach has also been (infrequently) applied to measure industrial water demands. For studying domestic water demands in rural developing countries, the contingent valuation method (discussed below as a method for valuing public environmental goods) has been successfully applied. Valuing Public Good Benefits of Water For public good type water uses, particularly for amenity and recreational valuation, other approaches have been refined in recent years. These can be divided into either "revealed preference" or "stated preference" approaches, both of which are based on surveys of actual or potential consumers. Of course, all the issues of sample survey planning-sample specification, questionnaire design, choice of survey technique-must be recognized and overcome. Relying on observations of actual expenditure choices made by consumers (revealing their preferences) the revealed preference methods infer net willingness to pay from the differences in expenditures observed with varying levels of environmental amenities or water supply. In the stated preference approach, (called the contingent valuation method) survey respondents are offered conditions simulating a hypothetical market in which they are asked to express willingness to pay for existing or potential environmental conditions (including water supply or quality) not registered on any market. The travel cost method is the most widely used example of the revealed preference methods. Suppose the absence of variation in fees for recreational and amenity sites precludes estimating directly the demand for such sites. If the cost of travel to a recreational site varies widely among consumers, and if these consumers respond to higher travel costs in the same way that they would respond to higher entrance fees, the analyst can derive a demand schedule for recreation at the site from the costs of travel. The hedonic pricing model is another revealed preference approach. The hedonic model applies to markets for goods which have several attributes which are recognized by purchasers, but the attributes cannot be unbundled when purchasing the good. The partial derivative of the hedonic price function with respect to the characteristic of interest yields a measure of the marginal value of that characteristic. The hedonic pricing method has been most frequently applied to the residential housing market, for analysis of real property (land) sales price data exhibiting differing but measurable environmental characteristics (e.g. varying water supplies or water qualities). Irrigation water supply represents another case to which hedonic pricing has been applied. Although the revealed preference approaches have the advantage of being based on actual consumer decisions, data for application to specific local planning conditions is often difficult to obtain. Turning to the stated preference method, valuation of instream flows and water quality benefits have been the main applications in water economics. The contingent valuation method has also been successfully applied to domestic water demand in rural areas of developing countries. The principal advantage of the technique is that it can potentially measure the economic benefits (or damages) of a wide assortment of beneficial (or adverse) effects in a way that is consistent with economic theory. A major plus is the xii

15 possibility of evaluating proposed, in addition to already available, goods or services. However, although a contingent value study can be an effective measurement tool where no other technique applies, some critics doubt the reliability of willingness to pay statements in situations where the respondent has no actual money at stake or is unfamiliar with the potential service being valued. To assure an accurate result, extreme care must go into design and conduct of the survey, particularly the definition of the good or service being valued. CVM studies, properly performed, require a significant research effort, well-trained staff and a budget to match. CONCLUSION Water valuation presents the economic analyst with a wide range of challenging issues and problems. Because water values tend to be quite site-specific, each case confronts its own unique issues, and typically requires its own original evaluation. Effective measuring of water values demands skill and rigor in application of all of the tools of the applied economists' trade. These tools include data collection, statistical analysis, optimization models and research reporting. The reader should recognize that estimating economic benefits for water-related decisions is seldom an easy task. Analysis of the demand side of water management decisions require as much specialized skill and training as is required by our colleagues from engineering and hydrology to perform their supply-side studies. However, those who are prepared to exercise the necessary skills and are given the time and resources to implement them effectively can derive conceptually consistent and empirically valid measures of the value of water, and thereby provide a valuable contribution to water resource management. xiii

16 Accounting price See shadow price. GLOSSARY Accounting stance The geographical area or political jurisdiction within which benefits and costs are accounted for (in a given economic evaluation). Averting behavior Actions people take to mitigate or avoid an external effect (such as water quality degradation). Reductions in the costs of averting behavior are a partial measure of the economic benefits of policies for reducing the externality. Change in Net Income Method A method for estimating the increment in producers' surplus associated with a change in the availability of a producer's or intermediate good. (This is a form of what is sometimes called "the Valuation of Productivity Change" or VPC approach to valuing intermediate goods). Consumer Surplus The excess in monetary value an individual would be willing to pay for a good over and above the total expenditures that would be made at a fixed price. Changes in consumer surplus associated with changes in quantity or quality of consumers' goods are the primary basis for measuring economic benefits in consumer goods contexts. Contingent Valuation A method of nonmarket valuation which asks individuals their values (in money terms) for specified changes in quantities or qualities of goods or services. Discount rate See Social rate of discount. Economic benefit A monetary measure of preference satisfaction or welfare improvement from some change in quantity or quality of a good or service. A person's welfare change is the maximum amount that person would be willing to pay to obtain that improvement. Extrinsic value Values that arise because things or acts are instruments for humankind for attaining other things of intrinsic value. (Also called instrumental values). Economic values are extrinsic. Hedonic valuation A revealed preference valuation approach which rests on the assumption that the price of some marketed good is a function of its different characteristics, and an implicit price exists for each of the characteristics. From a sample of closely similar marketed goods, implicit prices can be estimated with econometric techniques which reflect the value of the different characteristics of that good. Intermediate good See producers' good. Intrinsic value Assigned to things, actions or outcomes for their own sake, independent of means of providing or attaining other items or situations of value for humans. xiv

17 Private goods or services Those goods and services for which one person's consumption reduces the amount remaining for other consumers. (Also called rival goods). Can typically be allocated by price mechanisms. Water used by industries, agriculture or residential uses are examples of private goods. Producers' good A product or service used to make other goods or services (as contrasted with final consumption goods, used directly by consumers). Also called intermediate goods Producer surplus The amount a producer would pay for a production input minus the amount actually paid. Changes in producer surplus associated with changes in quantity or quality of producers' goods are the primary basis for measuring economic benefits in producers' goods contexts. Public goods Enjoyment of public goods by any number of individuals does not reduce the utility from the good by anyone else, public goods are not exchanged on markets. Quasi-public goods Up to a point, one individual's enjoyment of a quasi-public good does not affect enjoyment by others. However, beyond that point, congestion reduces the enjoyment of all users. Not usually exchanged on markets. Recreational and aesthetic uses of water are examples. Residual Method A method for measuring the value of an input used to produce intermediate goods. The approach approximates the marginal value product of a productive input, such as water, by subtracting all costs of production but one from the total value of output. The remaining (residual) value is assigned to the nonpriced input. Revealed preference methods Valuation methods, including travel cost and hedonic pricing, which are based on analysis of actual consumer transactions. Shadow price The value used in economic analysis when the market price is in some way an inadequate measure of economic value. Social rate of discount The shadow interest rate used by public agencies to discount future benefits and costs of public projects. Stated preference methods Nonmarket valuation techniques--such as contingent valuation--which ask individuals their values (in money terms) for specified changes in quantities or qualities of goods or services. Travel cost A revealed preference method of valuation which uses variations in the costs of travel to a recreational site as implicit prices for site usage in order to estimate the recreational demand for that site. Value added In any production unit, the difference between the value of output and the value of purchased inputs. Conventionally, labor and capital are treated as internal, rather than externally purchased inputs. Sometimes incorrectly used to impute the marginal value productivity or benefit of an increment of water. xv

18 Willingness to accept (WTA) A monetary measure of compensation defined as the minimum amount an individual would accept rather than experiencing some lesser amount or quality of a good or service. Willingness to pay (WTP) A monetary measure of the value an individual would pay to have a specified change in quantity or quality of a good or service. With-without principle The maxim that benefits (and costs) are to be measured as those strictly attributable to the project or policy (as contrasted to measuring changes before and after, which likely will include the effects of other factors). xvi

19 1. INTRODUCTION 1.1 THE NEED FOR SYNTHETIC ESTIMATES OF THE ECONOMIC BENEFITS OF WATER- RELATED DECISIONS Why a monograph on how to develop estimates of the economic benefits or economic value' of water? Estimates of the economic benefits relating to water management decisions are useful for several types of allocation decisions. Perhaps the most familiar is the contribution to decisions on investments on structural approaches to water management. Throughout the world, nations continue to make investments in water resources one of their most important component of public infrastructure budgets. Water-related investments - in irrigation, hydropower, urban and rural water supply, flood control and sanitation - have been designed to contribute to economic development and public welfare in many nations. Although most such investments were subjected to an economic evaluation to assure that they would represent an economical use of scarce water and capital, many earlier water resource investments have yielded less return than anticipated and have proven to have been based upon overoptimistic pre-project economic evaluations. Among the projects yielding disappointing results, many, it is clear, were evaluated with less than rigorous procedures. Economic evaluation is important because it aids in determining if people want proposed projects and estimating the degree to which they are willing to pay for benefits. In the prevalent situation of constrained public budgets, conceptually correct and empirically valid estimates of the economic contribution of water in each water using sector are essential for making economically sound investment decisions. Another class of decisions in which economic values of water are useful are those evaluating nonstructural or policy options. For example, as demands for fresh water grow against the finite world supply, estimates of the economic value of water are useful in the context of optimal allocation of water between and among water-using purposes and sectors. Water users will not be able to obtain all of the water they might possibly use. Sharing of the limited supply is a central issue of water management. In the context of water management, decision-makers in many nations face many other questions that invite economic evaluation, such as the following: How much water should be allocated to the agricultural sector for irrigated food production versus how much to cities with their household and industrial needs? How are needs to develop added food supplies to be balanced with the wish to preserve watercourses or wetlands for fish and wildlife habitat? How are wants for hydroelectric power generation and other instream uses to be balanced 'The term "value" takes on a narrow meaning in economics, referring to money measures of changes in economic welfare (Freeman, 1993, p.7). "Economic benefit" and "economic value" will be used interchangeably in this monograph to refer to positive welfare changes resulting from investment projects or policy initiatives. I

20 against demands for water from cities and farms? Each of the above cases are examples of the issue of optimal intersectoral allocation Several other nonstructural water policy problems for which water valuations are useful come to mind. These include: how much ground water should be pumped now and how much should be saved for future needs? How much ground water versus how much surface water should be withdrawn to meet current water demands? And, how much treatment to apply to wastes discharged into watercourses? Considering another dimension-that related to finance and cost recovery-how much can beneficiaries afford to pay for water supplies? For each of these issues, estimates of the net economic contribution of the water resource are important for water policy decisions. The reader will find a common theme running through the above survey of water allocation issues. Each of these are water management problems which involve choices as to how water should be combined with other resources so as to obtain the most public return from scarce resources. Included among the issues are the classic microeconomic resource allocation issues (Varian, 1993): how much of each input to use in production; how to proportion inputs in a production process; which products and how much of each to produce with scarce inputs; and how to allocate use of resources and consumption of goods and services between the present and future uses. Therefore, these issues can be usefully cast as resource allocation problems and can be best understood within an economic framework. It is a truism of applied policy analysis that "decisions imply valuation". Rational decision-making presupposes the forecasting of consequences, and assignment of values to these consequences. Because of the limited role played by market forces in the allocation of water, market prices upon which to base water-related resource allocation decisions are seldom available. In the jargon of the economist, accounting or shadow prices reflecting the value of water must be developed in their place. Economists have in recent decades developed a number of techniques for measuring the economic values or benefits associated with nonmarket allocation in the subject matter areas relating to the environment and natural resources. These techniques call for a wedding of economic theory and applied economic practice. The theoretical foundations of nonmarket economic valuation of environmental resources have come to be well developed (see, e.g. Freeman, 1993; and Braden and Kolstad, 1991). Progress with methods for estimating economic benefits in actual cases is also well advanced. Although much of the applied resource valuation literature has dealt with water resources in one or another of its many ramifications, there is no single publication which brings all these disparate methodologies together under one cover. Moreover, although many of the resource valuation techniques, particularly on the topic of environmental quality, have been subject to critical scrutiny and testing, some areas of water valuation have received less attention. Particularly for the intermediate or producers' goods derived from water - such as crop irrigation, hydroelectric power and industrial water use - procedures for empirical applications of valuation techniques appear to be less developed and seem to have received less application and critical challenge. 2

21 1.2 THE DISTINCTIVE NATURE OF WATER SUPPLY AND DEMAND A number of special characteristics distinguish water from most other resources or commodities, and pose significant challenges for the design and selection of water allocation and management institutions. On the physical side, water is usually a liquid. This trait makes it mobile: water tends to flow, evaporate and seep as it moves through the hydrologic cycle. Mobility presents problems in identifying and measuring specific units of the resource. Water supplies tend, due to natural climatic fluctuations, to be variable, so that the risks of shortage and of excess are among the major problems of water management. Water, due to its physical nature, and for other reasons, is what economists call a "high-exclusion cost" resource (Schmid, 1989). implying that the exclusive property rights which are the basis of a market or exchange economy are relatively difficult and expensive to establish and enforce. Frequently, then, property rights in water are incomplete or, more likely, absent. Turning to the demand side, humankind obtains many types of values and benefits from water. Because each of the different benefit types usually call for specialized evaluation and management approaches, it will be useful to group the types of water-related economic values into several classes. These are: (a) commodity benefits, (b) public and private aesthetic and recreational values; (c) waste assimilation benefits; and (d) disbenefits or damages. Each of these categories clearly involve economic considerations, because they are characterized by increasing scarcity and the associated problems of allocating resources among competing uses to maximize economic value. Whether certain other values associated with water, such as intrinsic values associated with endangered species preservation, ecosystem preservation and certain socio-cultural issues of rights to water, can be measured within the economic framework remains a matter of debate. We do not attempt to resolve that issue here. To consider water demand more closely, note that the economic characteristics of water demand vary across the continuum from rival to nonrival goods (Randall, 1987). A good or service is said to be rival in consumption, if one person's uses in some sense preclude or prevent uses by other individuals or businesses. Goods that are rival in consumption are the types that are amenable to supply and allocation by market or quasimarket processes, and are often called private goods. The opposite end of the continuum is occupied by goods that are nonrival in consumption, meaning that one person's use does not preclude enjoyment by others. Goods that are non-rival are often called public or collective goods. Because non-payers cannot be easily excluded, private firms will not find it profitable to supply public goods. Water for agricultural, residential or industrial uses tends toward the rival end, while the aesthetic value of a beautiful lake or stream is non-rival. The significance of non-rivalry can be better understood by noting its association with high exclusion costs. Exclusion cost refers to the resources required to keep those not entitled from using the good or service- Water is frequently a high-exclusion-cost good because of its physical nature noted above when the service exists for one user, it is difficult to exclude others. In such cases, it is hard to limit the use of the good to those 3

22 who have helped pay for its costs of production. (The refusal of some beneficiaries to pay their share of the provision of a public good from whose benefits they cannot be excluded is called thefiree rider problem. To circumvent the problem, public goods must normally be financed by general taxes rather than by specific charges.) The commodity benefits-the first type of benefit mentioned above-are those derived from personal drinking, cooking and sanitation, and those contributing to productive activities on farms and in businesses and industries. What are here called commodity values are distinguished by the fact of being rival in use, meaning that one person's use of a unit of water necessarily precludes use by others of that unit. Commodity uses tend to be private goods or services. Continuing with the discussion of commodity-type uses, some additional distinctions will be helpful. Those types of human uses of water, which normally take place away from the natural hydrologic system, may also be called withdrawal (or offstream) uses. Since withdrawal uses typically involve at least partial depletion or consumption (e.g., from evaporation and/or transpiration), they may further be distinguished as consumptive uses. Other types of economic commodity values associated with water may not require it to leave the natural hydrologic system. This group may be labeled instream water uses; hydroelectric power generation and waterways transportation being important examples. Since instream use often involve little or no physical loss, they are also sometimes called non-consumptive uses. (Although instream uses do not "consume" much water, in the sense of evaporating it to the atmosphere, they do often require a change in the time and/or place of availability- as is the case with water stored for hydropower generation-and therefore exhibit some aspects of the rivalness of a private good. ) The economic benefits from water for recreation, aesthetics, and fish and wildlife habitat are a second group or type of value of water. Benefits in this class are also closer to the nonrival end of the spectrum. Although aesthetics and recreation were sometimes viewed as nonessential goods inappropriate for public concern, as incomes and leisure time grow, these types of benefits are increasingly important. The populace of developed countries more and more often choose water bodies for recreational activities. In developing nations, water-based recreational activities are becoming more important for their own citizens, and also often provide a basis for attracting the tourist trade. As is waste assimilation, recreational and aesthetic values are also nearer the public good end of the spectrum. Enjoyment of an attractive water body does not necessarily deny similar enjoyment to others. (However, congestion at uniquely attractive sites, such as waterfalls or mountain lakes, may adversely affect total enjoyment of the resource.) Significant instream values also are found as habitat for wildlife and fish forms a basis for sporting activities. The economic benefits of waste disposal is a third general class of economic benefits of water use. Bodies of water are considered as a sink for carrying away a wide range of residuals from processes of human production and consumption. Water resources and used for disposal of wastes, diluting them, and for some substances, aid in processing wastes into less undesirable form. They are therefore significant for what is called their "assimilative capacity". The assimilative capacity of water is closer to being a public or collective (rather than private) value, because of the difficulty in excluding dischargers from utilizing these services. 4

23 Dis-values (also called damages or negative benefits) of water represent an important related classification. Examples are found in connection with evaluations of floodplain and water quality management. Flood waters or excesses of pollutants reduce value. Conversely, reduction of disbenefits increase human welfare. In such cases mitigation policies may be assessed by valuing the reduction in damages. Nonuse values are also an important consideration in water allocation, and for the economic valuation of water. It is observed, in addition to valuing the commodity benefits of water use, that people are willing to pay for environmental services they might neither use nor experience. Nonuse values are benefits received from knowing that a good exists, even though the individual may not ever directly experience the good. Voluntary contributions toward preserving an endangered fish species represent an example. Many resource economists argue that nonuse values should be added to use values so as to more accurately measure total environmental values. Because of differing conceptual frameworks, an additional useful distinction is between intermediate goods and final consumption goods. Intermediate goods (also called producers' goods) are employed to make final products (to be eventually used by consumers). Intermediate goods represent the largest class of offstream uses of water by humankind. For example, water for crop irrigation, the largest single consumer of water in the world, is an intermediate good; cotton or maize grown under irrigated conditions are destined to eventually be further processed to become clothing or food. Industrial processing and hydroelectric power generation are other intermediate uses of water. Consumption goods are those providing human satisfactions. Residential water is an example of a final consumption good from the private (rival) good classification, while recreation and amenity services provide nonrival final consumption values. The importance of this distinction between intermediate and consumer goods is that the economic theory of a profit-maximizing producer provides the conceptual framework for the valuation of intermediate goods, while the theory of the individual consumer is the basis for valuing consumer goods. Yet another useful distinction is between real and pecuniary economic effects. Real effects are actual changes in quantities of goods and services available, or changes in the amount of resources used. Real effects are positive or negative changes in welfare. Real effects are further subdivided into direct and indirect effects. Direct economic effects of water projects or policies are those which accrue to the intended beneficiaries; those that can be captured, priced or sold by the project entity, or -in the case of costs-which must be paid for. Indirect or external effects are those affecting third parties uncompensated side effects. Economists classify external effects as either technological or pecuniary. Technological externalities are real changes in production or consumption opportunities available to third parties, and generally involve some physical or technical linkage among the parties (such as degraded water quality). This type of externality represents a change in welfare, and should be reflected in evaluation of the economic efficiency effects of policies or projects. Pecuniary impacts (often referred to as secondary economic impacts in the water planning literature) are those reflected in changes in incomes or prices, (such as effected by increased purchases of goods and services in a regional economy). Secondary economic impacts typically represent income distribution impacts. 5

24 From the larger perspective of nation or state, secondary impacts registered on a specific locality are likely to offset by similar, but more difficult to isolate, effects on income of opposite sign elsewhere. Economic convention therefore suggests that secondary impacts not be taken into account in economic evaluations, or only in special cases. This monograph will limit its focus to measuring real economic impacts, both direct and indirect, and will ignore secondary (pecuniary) effects. 1.3 OBJECTIVE, SCOPE AND PLAN OF THE MONOGRAPH This monograph is designed to provide a review, exposition and critical assessment of the concepts and methods for estimating economic benefits of water investment and allocation decisions involving water. It is intended for field practitioners possessing some (preferably postgraduate) training in applied economics who face the task of evaluating water-related decisions. The aim is to provide such analysts with a practical, yet theoretically sound procedural handbook to help guide their analyses. This monograph owes an intellectual debt to collaborators in preparing an earlier document (Young and Gray, et al, 1972). The earlier document limited its focus on a developed country situation, and as a U.S. Government contract report, had relatively limited circulation and is now outdated in many respects. Two other earlier works on the general topic are Kindler and Russell (1984) and Gibbons (1986). The former focused primarily on water demand (rather on the subsequent step of deriving benefit measures for project and policy planning), while the latter emphasized empirical estimates, rather than methodology. Conceptually sound and empirically accurate estimates of the economic benefits of water applicable to a specific policy proposal is most often a task demanding of more time, resources and technical skills than is generally recognized by nonspecialists. It requires familiarity with both positive and normative microeconomic theory and the tools of applied quantitative economics. Water valuation rests on the normative framework of neoclassical welfare economics, as it is a particular application of applied benefit-cost analysis (Just, Hueth and Schmitz, 1982; Smith, 1987). Positive microeconomics provides the conceptual framework for the analysis of behavior of producers and consumers (Varian, 1993). Additionally, any of the numerous techniques from the quantitative toolkit of applied microeconomics may be used. The applicable techniques range from survey research to econometrics to optimization modeling. This monograph does not aspire to reflect all of the ramifications of the various concepts and analytical techniques which can be brought to bear on estimating economic benefits of water related investments or policies However, the intended scope is to provide enough conceptual framework and methodological detail for a persistent reader to understand the various potential applications and the strengths and limitations of each, together with sufficient references to the relevant literature to aid the user to find further specifics, if needed. The remainder of the monograph is organized as follows. First, to set the stage, in Chapter 2 we briefly review the basic conceptual and analytic issues underlying economic valuation of water. However, numerous excellent analyses emphasizing the conceptual 6

25 bases of nonmarket valuation are available (e.g. Freeman, 1993; Braden and Kolstad, 1991; Johannson, 1987). Here, the major emphasis is on applications. The various techniques for nonmarket evaluation of water are then described and their advantages and limitations noted in Chapter 3. The remainder of the main body of the document takes up the practice of valuation as applied to individual water use sectors. Chapter 4 addresses valuation of intermediate goods, such as crop irrigation, industrial and hydropower generation. Chapter 5 takes up the task of evaluating water in its role as a private consumers' good, for domestic water supply. In Chapter 6, issues of benefit measurement for instream public goods, including values of water-based recreation, water quality enhancement and flood risk reduction are briefly discussed. 7

26 2. CONCEPTUAL FRAMEWORK AND SPECIAL PROBLEMS IN VALUING WATER This chapter summarizes the conceptual framework for economic valuation of nonmarket goods and services as applied to water resources. It begins by reviewing some of the basic concepts and definitions used in measuring economic value or benefits of public water projects or policies. The main focus is on changes in water supply, but changes in quantity and reliability are also addressed. The chapter concludes with a discussion of some issues in valuation relatively unique to the water resource. 2.1 INTRODUCTORY COMMENTS Resources have economic value or yield benefits whenever users would willingly pay a price for them rather than do without, i.e., whenever resources are scarce. Effective market operation results in a set of market values (prices) which serve to allocate resources and commodities in a manner consistent with the objectives of producers and consumers. In many parts of the world. water has been plentiful enough that it could be regarded as a free good, and until recently, institutional arrangements for managing water scarcity have not been of serious concern. When markets are absent or do not operate effectively, economic evaluation of resource allocation decisions requires that some means of estimating resource value be found. In either case, resource value is measured in the context of a specific objective or set of objectives. The value of the resource reflects its contribution to the objectives. In water resources, governments have identified several objectives which may be relevant: enhancing national economic development, enhancing regional economic development; enhancing environmental quality; and enhancing social well-being (U.S. Water Resources Council, 1979; 1983; Organization for Economic Cooperation and Development, 1985). 2.2 ECONOMIC VALUE VERSUS OTHER CONCEPTS OF VALUE The economic approach is not the only way to assign values to natural and environmental resources. Broadly speaking, values can be termed extrinsic or intrinsic, both of which are relevant for water and environmental policy. The distinction rests on whether the basis for valuation derives from consequences for human welfare. Extrinsic (sometimes also called instrumental) values are those that arise because things or acts are instruments for humankind for attaining other things of intrinsic value. As an example, water resources may be valued (instrumentally) for their contribution to human health, welfare or satisfactions. Intrinsic values, in contrast, are assigned to things, actions or outcomes for their own sake, independent of means of providing or attaining other items 8

27 or situations of value for humans (Anderson, 1993: ). For example, people often value environmental resources in ways other than from their use or consumption by humans; the public wishes to preserve endangered species or protect delicate ecosystems, without consideration of whether these offer immediate human utility. It is important to recognize that both approaches to valuation are legitimately applied to environmental and resource policy (Pearce, 1993; 13-15). However, the prevailing-though not unanimous view of philosophers- is that neither extrinsic nor intrinsic values are necessarily absolute. When values conflict, as they often do, a dilemma arises. In such cases, the only apparent solution is to make a judgment of how to compromise the competing goals (MacLean, 1993). Morgan and Henrion (1990, p 27) describe a widelyused approach-called the approved process approach, which roughly speaking, requires all relevant parties to observe a specified set of procedures or observe a concept of due process to estimate a policy's impacts on relevant measures of value. Any decision reached after an appropriate authority balances the competing values under the specified procedures is deemed acceptable. Standard water planning manuals-both the U.S. Water Resources Council's Principles and Guidelines (1983) and the Organization for Economic Co-operation and Development's Management of Water Projects (1985)-although neither acknowledge the underlying philosophical premises-appear to reflect an approved process approach. Both manuals call for a determination of environmental impacts (intrinsic values) to be balanced against human (economic and social) welfare (extrinsic value) considerations. Both manuals emphasize display of impacts - the ultimate resolution or balancing of conflicting values is assumed to be made at the political, rather than the technocratic level. The economic values discussed in this monograph are extrinsic (instrumental), in that they reflect people's assessment of contributions or decrements to human welfare. These economic benefits will be appropriate to either a stand-alone economic analysis or as part of a more general multi-objective or approved process approach. 2.3 ECONOMIC CRITERIA FOR RESOURCE ALLOCATION AND VALUATION Although the objectives of improving the distribution of income, enhancing environmental quality, and attaining other nonmarket goals are important, the analysis in this monograph pertains exclusively to the objective of economic efficiency in the development and allocation of the water resource. There are two major reasons for this: first, under conditions of increasing scarcity and growing competition among water users, economic efficiency remains an important social objective and efficiency values have viable meaning in resolving conflicts; second, efficiency values provide a valuable means of assessing the opportunity costs of pursuing alternative objectives. 9

28 2.3.1 The Pareto Principle and Economic Efficiency Economic efficiency may be defined as an organization of production and consumption such that all unambiguous possibilities for increasing economic well-being have been exhausted. Stated somewhat differently, economic efficiency is an allocation of resources such that no further reallocation is possible which would provide gains in production or consumer satisfaction to some firms or individuals without simultaneously imposing losses on others. This definition of economic efficiency (termed Pareto optimality) is satisfied in a perfectly functioning competitive economy. Abstracting from the mathematical elegance found in textbook expositions (e.g. Just, Hueth and Schmitz, 1982) and abstracting further from the time consideration in outputs and inputs of economic activities, Pareto optimality can be expressed quite simply in terms of the attainment of: (1) economic efficiency in production of goods and services; (2) economic efficiency in distribution of goods and services; and (3) resource allocation in a manner consistent with consumer preferences. Pareto efficiency is said to occur when the marginal benefits of using a good or service are equal to the marginal cost of supplying the good. Pareto optimality rests on several central value judgments (Maler, 1985). The first of these is the judgment that individual preferences count; the economic welfare of society is based on the economic welfare of its individual citizens. Second, the individual is the best judge of his/her own well being. The third, highly restrictive, value judgment is that a change which makes everybody better off with no one becoming worse off constitutes a positive change in total welfare From Theory to Practice Translating from the welfare economics theory to benefit-cost practice requires further steps. Because in a complex modern society, few policy changes which improve welfare for many would avoid lowering welfare of some individuals, few proposed changes would meet the strict Paretian standard of making no one worse off. However, welfare theorists circumvented this problem with the compensation test: if gainers could compensate losers and still be better off, the change would be judged an improvement. In practice, compensation is often impracticable; identifying and compensating all adversely affected parties is expensive and time-consuming. Hence, the compensation test becomes a test for a potential Pareto improvement. If gainers could in principle compensate losers, the change is deemed acceptable, whether or not the compensation actually takes place. Also, rather than evaluating all possible allocations in a continuous function framework, benefit-cost analysis typically examines fairly large discrete increments of change to assess whether the move is in the direction of Pareto efficiency (Smith, 1986). An action which generates incremental benefits in excess of incremental costs is termed Paretosuperior, because it leads to a condition superior to the status quo ante. Following Smith (1986), Figure 2.1 illustrates the comparison of Pareto efficiency and benefit-cost criteria. The curve denoted B(W) is a representation of aggregate benefits (i.e. consumer or producer surplus) of alternative levels of water services (W), while C(W) 10

29 represents the associated aggregate costs. These curves measure social welfare or aggregate utility and cost. Their general forms reflect the conventional assumption that benefits increase at a decreasing rate with increased output and costs increase at an increasing rate. The Pareto-efficient solution is at W* -the maximum vertical distance between B(W) and C(W). At W* the marginal benefits equal the marginal costs. However, rather than seeking a full optimum solution, benefit-cost analysis (BCA) in practice typically considers whether a change from given conditions would represent a desirable shift. In Figure 2. 1, such a change would be represented by moving from W 1 to W 2. The conventional BCA test compares the aggregate increment in benefits (GH in Figure 2.1) with aggregate incremental costs (EF). If incremental benefits exceed incremental costs, as they are drawn to do in the Figure, then the change is termed a Pareto improvement. Any act or policy judged a Pareto Improvement would be recommended as preferable to the existing situation. 2.4 ECONOMIC VALUATION IN THE ABSENCE OF MARKET PRICES Water management policies can have widespread effects on the quantity of water available, its quality, and the timing and location of supplies for both in- and off-stream uses. In general, these impacts have an economic dimension, either positive or negative, which must be taken into account in policy formation. Specifically, the decision process (resolution of conflicts) requires the identification and comparison of the benefits and costs of water resource development and allocation among alternative and competing uses. Beneficial and adverse impacts to people are abstract and often ambiguous concepts. As noted earlier, mainstream economists treat values as extrinsic, and propose to measure impacts in terms of satisfaction of human preferences. To transform the concept of welfare into a single metric, the suggested measuring rod is that of money (Rhoads, 1985). A person's welfare change from some proposed improvement is measured as the maximum amount of money a person would be willing to forgo to obtain the improvement. Conversely, for a change which reduces welfare, the measure is the amount of compensation required to accept the change. The economic evaluation of projects or proposals is based on balancing the predicted beneficial against the adverse effects generated by the proposal. Benefits are the "good" or "desired" effects contributed by the proposal, while costs are the "bad" or "undesired" impacts. This balancing of costs against benefits is called benefit-cost analysis (BCA). (For detailed treatment of the overall approach to BCA-particularly as applied to environmental and natural resource problems- the reader is referred to the extensive literature in that field, e.g. Pearce, 1987; Randall, 1987; Schmid, 1989; Johansson. 1993; Pearce and Warford, 1993; Dinwiddy and Teal; 1996; Winpenny, 1996) 11

30 Fig 2.1 Comparison of Pareto-Efficiency and Benefit-Cost Criteria (After Smith, 1987) d B (w*) dw CM(W) 0 G d C (W*) dw / 3 WI W 2 W* In applied BCA, the terms benefit and cost are assigned a narrow technical economic interpretation. The prices used in BCA are interpreted as expressions of willingness to pay (WTP) for a particular good or service by individual consumers, producers or units of government. Direct benefits are willingness of beneficiaries to pay for project services or policy impacts. Direct costs are willingness to pay for the foregone alternatives, or to avoid any adverse effects. In what follows, changes in producer surplus and consumer surplus, respectively, are accepted as the pertinent measures of willingness to pay or to accept compensation. 12

31 2.4.1 The Need for Shadow (Accounting) Prices Howe (1971) has classified policy impacts into four categories which are paraphrased below: 1. Impacts for which market prices exist and market prices reflect scarcity values. 2. Impacts for which market prices may be observed, but such prices fail to accurately reflect true social values, but they can be adjusted to more accurately do so. 3. Impacts for which market prices do not exist, but it is possible to identify surrogate market prices. 4. Impacts for which market-prices or surrogate prices are not meaningful. The second and third cases are most typical in benefit-cost analysis for water resource planning; in these instances the prices employed are called shadow prices (or sometimes accounting prices). Benefits and costs must be expressed in monetary terms by applying the appropriate prices to each physical unit of input and product. Three types of estimates are employed. A primary source of the prices used for BCA are the result of observing the market activities. However, in the second type, (often the case in water planning) it is necessary to make adjustments to observed market prices (for example, when agricultural commodity prices are controlled by government regulation or when minimum wage rates are set above market clearing prices.) Finally, in many cases, it will be necessary to estimate prices which do not exist at all in any market (such as the value of water used for power generation). When either of the latter two cases occurs (adjusted or estimated prices) in cost-benefit analysis, the prices employed are called shadow prices Defining Shadow Prices: The Willingness to Pay Principle Whatever the source, the prices used in BCA are interpreted as expressions of willingness to pay or willingness to accept compensation for a particular good or service by individual consumers, producers or units of government. This presumption is obvious for market prices, since the equilibrium market price represents the willingness to pay at the margin of potential buyers of the good or service. For non-marketed goods, WTP also is the theoretical basis on which shadow prices are calculated. The assertion that willingness to pay should be the measure of value or cost follows from the principle that public policy should be based on the aggregation of individual preferences. Willingness to pay represents the total value of an increment of project output, i.e. the demand for that out- 13

32 put. 2 Willingness to accept compensation (WAC) is an important welfare measure in some contexts. WAC is the payment that would make an individual indifferent between having an improvement and foregoing the improvement while receiving the extra money. Alternatively, it is the minimum sum that an individual would require to forego a change that otherwise would be experienced. WTP and WAC may not be equal, because WTP is constrained by the individual's income. (See Freeman, 1993, and Hanemann, 1991 for further discussion). Therefore, benefits are defined as any positive effect, material or otherwise, for which identifiable impacted parties are willing to pay. Costs are the value of the opportunities foregone because of the commitment of resources to a project, or the willingness to pay to avoid detrimental effects. (Critics of certain applications of BCA from within the ranks of economists, observing that WTP is dependent on the existing distribution of income, caution against any unquestioning application of the technique for public investment decisions. Bromley, 1989, for example, seems to contend that WTP would be different enough under more egalitarian income distributions that benefit-cost tests might be significantly revised. Although some analysts have made efforts to adjust accounting prices for income distribution considerations, the author, along with the majority of practitioners is less concerned, and is content to leave redistribution to the political process.) Economic Surplus and Measures of Benefit Economists base the concept of economic value on a decision framework within which rational individuals make the best use of resources and opportunities. The framework assumes that the individual members of the economy react systematically to perceived changes in their situation. Such changes can include- in addition to the quantity and quality of the water resource of primary interest here-prices, costs, institutional constraints and incentives, income and wealth. Figure 2.2 illustrates the concepts of economic (producer's or consumer's) surplus under marketed commodity conditions. The curve denoted MBW' in figure 2.2 is a familiar demand curve, reflecting the maximum amount of the commodity w that consumers would be willing to take at alternative price levels. The demand curve slopes downward to the right, reflecting the desire for consumers to take more of the commodity w only as the price declines. The inverse demand curve can also be interpreted as the marginal willingness to pay for alternative quantities, so it is conventionally labeled in benefit-cost analysis, as in Figure 2.2, a marginal benefit (MB) function Consumers surplus is defined as the area above the price: it represents the difference between the maximum users would be willing to pay and what they would actually pay under a constant price per unit. The supply curves SI and S 2 represent a nonmarginal shift in supply functions, such as from a 2 Some authors, unfortunately, in addition to this broad meaning, use "willingness to pay" to refer to a specific type of nonmarket valuation study which directly questions people on their valuations for environmental changes. To avoid ambiguity, these specific techniques would best be identified by the name of the relevant elicitation process--e.g. "contingent valuation." 14

33 project which increases the supply of some productive factor, such as water for crop irrigation. Consumers enjoy two forms of gain: a decrease in unit price from PI to P 2 and an increase in available output (from w, to w 2 ). Producers also see a gain, from expanded output, but their price goes down. The area in Figure 2.2 circumscribed by the points PIABP 2 represents the gain in surplus enjoyed by consumers. With the change from w, to w 2, producers surplus changes from PIAD to P 2 BE, The net increase in economic surplus, the sum of gains and losses to producers and consumers, is DABE. Fig 2.2 Price and Quantity Effects and Change in Economic Surplus from Nonmarginal Shift in Supply of Water /SI /S P 1 D E MBW 0 W Quantity The economist reading the above paragraphs will note that the measures shown are for the ordinary Marshallian concept of demand and consumer surplus. More precise welfare measures, called Hicksian measures are often reported in the applied welfare economics literature (Just, Hueth and Schmitz, 1982). The Hicksian compensating version refers to the amount of compensation (received or paid) which would return the individual to their initial welfare position. The equivalent version refers to the amount of money that must 15

34 be paid to the consumer to make him/her as well off as they could have been after the change. Whether to aim for the Hicksian formulation depends on the individual case. Marshallian demand functions are sometimes easier to estimate. Moreover, when purchases of the good or service in question accounts for only a small part of the household budget, it has been shown that the Marshallian measure is often a quite close approximation to the Hicksian measure. (See Freeman, 1993, for a full analysis). For the case of water resources, which for the most part makes up a small fraction of consumers' budgets, the differences among the measures are probably smaller than the errors in estimating the functions econometrically, so the Marshallian approximation will usually be acceptable in the practical applications discussed in this monograph. Figure 2.3 portrays a case frequently applicable to nonmarket valuation applied to water resources (Randall, 1987; chap. 13). It represents an increase in the availability of a nonpriced water use from w, to w 2. Perfectly inelastic supply curves SI and S2 shift from wl to w 2. The curve MB,, as before, shows the downward sloping marginal benefit function. The area under MB, between w, ("without change") and w 2 ("with change") represents the economic surplus attributable to the changed water supply. This area is that bounded by the points w,abw 2. It is this area that the economic analyst is attempting to measure in nonmarket valuation of changes in water and environmental amenities. Note that the curve MB, can, in addition to representing consumer demand, also portray producers' surplus. In the latter interpretation, MB, is the producers' Marginal Value Product (MVP) function, the marginal net return to increasing level of input. (See Johannson, 1993, Section 5.1, for a formal derivation of these properties of producer's surplus). This interpretation is, in fact, more applicable to valuation purposes than the producers surplus depicted in figure 2.2, (i.e. the area above the supply curve S and below the price line). Also, in parallel with the Hicksian adjustment for income effects to consumer surplus measures, a corresponding adjustment for cost-minimizing allocation of other inputs or technology is appropriate for producer surplus measures (Johannson, 1993). To recapitulate, the economic value of an environmental resource is measured by the summation of many users' willingness to pay for the good or service in question. Willingness to pay is a monetary measure of the intensity of individual preferences. Therefore, we can say that economic valuation is the process of expressing preferences for beneficial effects or preferences against adverse effects of policy initiatives in a money metric. 16

35 Fig 2.3. Change in Economic Surplus from Nonmarginal Change in Water Supply s ~~S 2 0) 4.J > B B - MBt (D WI W2 Quantity Opportunity Costs: Measuring Foregone Benefits of Reduced Water Use Increasingly of interest are measures of opportunity costs of water resources. Opportunity costs are the benefits foregone when a scarce resource is used for one purpose instead of the next best alternative. When evaluating tradeoffs of proposed reallocations, one needs a measure of the benefits of the proposed new use as well as the reduction of benefits associated with reduced water use in the sector currently benefiting. Hence, opportunity costs are conceptualized as the reverse of incremental benefits. Returning to Figure 2.3, a measure of opportunity costs would be the area under MIBW from, this time, w 2 to w,. This is the same area as described before, that in Figure 2.3 bounded by the points w,abw 2. (Randall, 1987, Figure 13 5 conceptualizes this point more elegantly in a framework jointly accounting for increments or decrements of natural resource use.) 17

36 2.5 SOME SPECIFIC CASES OF ECONOMIC EVALUATION OF WATER RESOURCE ISSUES As noted in the Introduction, in a river basin management context, the principal opportunities for economic welfare enhancement, and hence, the need for measures of water value, are, first, investments in capturing, storing, delivering and treating new water supplies and second, reallocation among water-using sectors. Other examples where marginal values of water might be useful include: optimal ground water basin policy (e.g. Young, 1992; Provencher and Burt, 1994) and pricing and cost recovery for investments in water supply systems. Of most interest are the cases of investment and reallocation decisions, discussed below in more detail Evaluating Investments in Additional Water Supplies Consider an investment in a multiple purpose water project which is being assessed for its economic feasibility. The economic feasibility criterion can be written: PVSNB = t [Yi (Bi,)/(1+r) t ]-Zt[(Ct)I(1+r)'] - F, [(Dj,)/(I+r) t ] (2.1) where: PVNB = Present Value of Net Benefits; B;, represents Incremental Benefit (willingness to pay) for incremental water use or availability in sector i in year t; Ct is capital and operating costs in year t; Djt represents incremental project-induced disbenefit (foregone benefits or external costs) to sector j in year t; and r is the discount (interest) rate. The PPI (Potential Pareto Improvement) hypothesis to be tested is: Is PVNB > 0? (Of course, the PPI test can be also expressed in the alternative, but largely equivalent forms of Benefit-Cost Ratios or Internal Rates of Return. See. e.g. Gittinger, 1983, for discussion). In implementing this test, economic valuation will be required for the terms B 11 and Djt Evaluating Proposals to Reallocate Water Among Sectors Another likely welfare improvement opportunity is for reallocating water among use sectors. The analytic question is: can a reallocation from sector i to sector j yield 18

37 incremental gains to sector j in excess of the foregone benefits in sector i? The hypothesis (for a Potential Pareto Improvement) to be tested is Is: YtIM[Bj/(l+r)' t # X,MBj /(l+r)'? (2.2) (for if they were equal, no gains from reallocation would be possible.) In applied cases, the hypothesis of suboptimal allocation is tested for specific proposals for reallocation. Consider a proposal to reallocate water from agriculture to municipal uses. Indirect impacts are expected on the hydropower sector. The PPI test can be expressed by developing measurements for two conditions (Young, 1986): The first condition is that the benefits (both direct and indirect) to the municipal sector exceed the sum of (foregone direct benefits to the selling sector plus foregone indirect benefits to the selling sector plus foregone indirect benefits to the hydropower sector. Condition 1 can be written (assuming all benefit and cost expression are in present value terms, employing a consistent planning period and price level): DB+IB>FDB+FIB + FIB+TC+CC (2.3) where: DB: IB: FDB: FIB TC: CC: Direct Economic Benefit (Value) to receiving sector Economic Benefit to Indirectly affected sector(s) Foregone Direct Benefit (value foregone) in source sector Foregone Benefit in indirectly affected sector(s) Transactions costs (for information, contracting and enforcement) Conveyance and Storage Costs A further condition is that the Direct Foregone Benefits in irrigated agriculture be the least-cost source of water for the purchasing sector: FDB+FIB+TC+CC<AC (2.4) That is: Condition 2 asserts that the sum of direct and indirect foregone economic benefits and the transactions and conveyance costs should be less than the cost of the next best alternative water source. 19

38 Economic analysis of both issues-as well as the other resource allocation and cost recovery problems mentioned in the Introduction-require the estimation of incremental or marginal benefits of changes in water supply or use. The overall task of our study is to critically examine methods for estimating the various manifestations of incremental benefits. This discussion has focused on measuring benefits of increments or decrements of water supply. To this point, the analysis has abstracted from two other dimension of water supply-water quality and supply reliability. These are taken up in the next subsections. 2.6 THE BENEFITS OF IMPROVED WATER QUALITY The quality of water, of course, also influences its economic value. Water in natural environments is never perfectly pure. Humankind uses water bodies as sinks for disposal of numerous wastes from production and consumption activities. The extent to which micro-organisms, and dissolved or suspended constituents are present varies greatly, and in sufficiently high concentrations can affect health, and reduce aesthetic values and productivity. Therefore, the content of pollutants, or conversely, the degree to which the water is treated for various uses is important in determining its economic value. Estimating benefits of improved water quality raises some complex and challenging issues. For the important cases of degradable effluents-those which are transformed after discharge into receiving waters-the detrimental effects depend on the nature of downstream water uses, the distance downstream, temperature, rates of flow and the quality of receiving waters. Willingness to pay for water quality improvement is usually assumed to reflect damages to subsequent water users. The damage function is a measure of the effect of the concentration of pollutants on the utility or costs of receiving entities. Benefits are the damages avoided from a given project or regulative policy. The framework for conceptualizing the benefits of water quality improvement can be readily derived by extending the model developed earlier for increments of water supply. All other factors (prices, incomes, technologies, etc.) held constant, an improvement in quality of water for either producers or consumers will shift the demand or marginal benefit curves to the right. The increment in producers' or consumers' surplus accruing to the change will be the appropriate measure of benefits of an improvement in water quality (See Spulber and Sabbaghi, 1994, Chapter 2 for a rigorous exposition). A related example responds to the need for measuring economic damages from releases of harmful materials into public water bodies. This issue has increasingly come into prominence in the U.S. in response to the Comprehensive Envircnmental Response, Compensation and Liability Act (CERCLA) of 1980 (see e.g. Kopp and Smith, 1993). 20

39 2.7 BENEFITS OF IMPROVED WATER-SUPPLY RELIABILITY The degree of certainty with which supplies are available, in addition to its quantity and quality, is another important factor influencing the willingness to pay for water. Domestic, industrial and agricultural water demanders all place a higher value on reliable water supplies than on supplies with high risk of availability. At least two cases can be envisioned for which the potential for changed reliability might have value The major source of water supply unreliability comes from normal hydrologic risk; reflecting the inevitable swings in precipitation and runoff (For individual users, hydrologic variation may be exacerbated by the institutional arrangements for sharing shortages. Where the rule for allocating shortages is a priority-first in time/first in right-system, high priority users may be little affected while low priority classes experience more than proportional fluctuation.) Another, less common, problem is the short-term lack of reliability of water supply systems, due to either inadequate capacity or to breakdown. Some third world cities, for example, lack sufficient capacity to be able to deliver water regularly to all customers on demand. Limited available water supplies may be rationed by rotating among different geographic sectors of the city's system. In such cases, even customers with piped residential connections are unable to obtain water on demand throughout the full 24-hour day, or even be unable to obtain some water every day (Nickum and Easter, 1994). Increasing reliability comes at increasing costs, so tradeoffs are necessary between cost and risk. Conventional technical risk analysis as applied to water supply reliability selects a risk level roughly reflecting the potential severity of adverse effects, and designs projects to satisfy the selected degree of risk (Renn, 1992). Reliability standards typically vary among use classes: for reasons of health, sanitation and implicitly, willingness to pay, water supply reliability is usually set higher for domestic supplies than for irrigation. The technical approach treats all affected areas and parties equitably, but it ignores economic efficiency considerations. Under technical reliability standards, investments to improve reliability may not subjected to systematic comparison of costs of improved reliability with the expected losses averted. Therefore, large expenditures may sometimes be made which have little prospect for a corresponding reduction of damages. In contrast, the economic approach goes beyond the identification of the probability of some adverse event to the measurement of the disutility of such events to humans. Howe and Smith (1994) developed a model for assessing reliability and apply it to the case of municipal water supply. Of interest here is how they formulate a function reflecting the economic benefits of reliability to compare with costs of achieving reliability. They defined the "Standard Annual Shortage Event" (SASE) as a drought of sufficient severity and duration that certain specified restrictions on water use would be put in place. (Howe and Smith's case study was for cities in the semi-arid western United States, so the hypothesized drought-induced restrictions were on summer, outdoor water usage for lawns and gardens). Here, the discussion abstracts from the optimization model formulated by Howe and Smith to focus on the marginal economic benefit of improved reliability. System reliability, R, is defined in terms of probability (P) of occurrence of the SASE. 21

40 R(SASE) 1_ - P(SASE) (2.5) Next, a loss function L(SASE), is introduced, which represents the reduction in economic value accruing if the SASE were to occur. The desired economic measure, the marginal benefit of improved reliability is given by the incremental reduction in expected losses de(l)/dr (2.6) Valuing reliability has received relatively little attention. The question of how to estimate benefits of improved reliability is taken up under discussions of the specific water use categories. 2.8 OTHER CONCEPTUAL ISSUES IN WATER VALUATION A number of additional concepts are important for applied economic evaluation in water resource management. They include the with atnd without principle, the accounting stance and the distinctions among total, average and marginal values The With-Without Principle The with and without principle holds that policy appraisal should contrast the "state of the world" as it would be with the policy to the "state of the world" as it would be without the policy. An important implication of the principle is that project evaluation is not adequately accomplished by comparing conditions before the project with conditions after its implementation. Many changes in the world from "before" to "after" would have occurred without the project, so such effects should not be credited or charged to the project. The With-Without principle directs the analyst to measure the impacts according to the status of the economy with the public intervention as compared to without the intervention. The intent is to identify only the impacts that are clearly associated with the project or program, and not include as impacts any changes in the economy which would have occurred even without the project. Therefore, regional growth which would be due to private sector investment, or to other public projects should not be included in project impact measures. Project evaluation which measures impacts by comparing before with after the intervention are likely to overstate project impacts. 22

41 2.8.2 Accounting Stance The accounting stance is defined as the geographical area or political subdivision within which benefits, costs or other impacts are counted in a CBA. A project may have impacts which are confined to a local area, or they may extend to the nation or even internationally. For example, an irrigation project might generate benefits in a local area. The conventional direct costs of construction and operating would normally be met by water users (or taxpayers) in the project area. Other costs, particularly indirect or external costs, such as foregone electric power generation or lower water quality imposed on downstream water users, will accrue well beyond the borders of the area benefited, but need to be accounted for in a full economic evaluation. Indirect benefits outside the project region can also occur. For example, interception of flood waters by irrigation or power reservoirs may yield benefits far downstream. Ideally, the accounting stance should be as encompassing as possible; impacts should be accounted for no matter how far away or in what political jurisdiction they may occur. For example, indirect costs of water projects in upstream nations adversely affecting downstream neighbors (such as by removing water) or indirect benefits (such as by intercepting flood waters) should be assessed. However, in practice, the choice of accounting stance must be made on pragmatic grounds, balancing the gains in accuracy against the increased costs of spreading a wider net. Most national planning agencies suggest a national accounting stance wherever possible. 2.9 SPECIAL PROBLEMS ASSOCIATED WITH NONMARKET VALUATION OF WATER A number of additional economic and physical characteristics give distinctive problems to the task of water valuation. The general point is that there is no single economic value of water. What is being measured is the welfare change associated with some policy-induced change in the attributes of the commodity. It is important to keep clear what are the specific attributes of the situation and decision in question. A number of these issues in are discussed in this section Long-run versus Short-run Values Because policy decisions relating to water entail a range of cases, from major longlived capital investments to one-off allocations in the face of immediate events such as droughts, it is often important to distinguish carefully between long run and short run values. The distinction relates to the degree of fixity of certain inputs, and is particularly important for cases in which water is a producers' or intermediate good, such as in irrigation, industry and hydropower. Anticipating the model to be developed in more detail in the next Chapter, consider a rational producer's willingness to pay for water. In the short run, where some inputs are fixed, the estimate of the net increase in the value of output can ignore as sunk 23

42 the cost of the fixed inputs. In the long run, where input costs must all be covered, these costs cannot be ignored. Therefore, we would expect that for the same site and production processes, values estimated for short-run contexts will be larger than values for the long run. Similarly, domestic water users exhibit different responses in the short versus the long run. Price elasticity of demand is less (in absolute value) in the short run when decisions are constrained, than in the longer run decision context, when adjustments to shortages are possible. Accordingly, willingness to pay in the short run planning situation is typically higher than in the long run case. Most public water policy decisions involve situations where the long run context is appropriate. Failure to observe this distinction has caused many nonspecialists to erroneously attribute too high a value to intermediate water uses. However, important cases occur-such as drought planning-where short run values are appropriate The Value of Water Relative to Other Commodities Water falls among the commodities which economists call "bulky." This means that the economic value per unit weight or volume of water tends to be relatively low. (For example, retail prices for water delivered to households are typically in the range of US per liter or about US$1. to $1.5 per ton, much less than other important liquids important in contemporary life, such as petrol (gasoline), milk or soft drinks. In crop irrigation, much of the water applied may yield direct economic values-profit after production costs-of less than US$0.04 per ton.) Bulkv commodities tend to exhibit high costs of transportation per unit volume, so that costs of transporting them become an increasingly important part of the total cost of supply. In the case of water, this point implies that water values are often highly site-specific, and moving the commodity from one place to another frequently may become uneconomical due to conveyance costs Commensurability of Place, Form and Time Marketed economic commodities are priced according to spatial, quality and temporal attributes, and shadow pricing of water should follow similar rules. For example, another economically important liquid, petroleum, is always priced in terms of grade, location and date of delivery. A look at a daily newspaper's business pages reveals that prices for crude oil at the point of production are less than the cost per unit volume of refined gasoline (petrol) in bulk at some specified distribution point, which is in turn much lower than the price of gasoline at the local retail station. These considerations lead to a need for analysts engaged in comparative water valuation exercises to be careful to assure that the chosen measure of water value is commensurable in terms of a common denominator of place, form and time. Consider first the aspect of location. Because of its low value at the margin, capital and energy costs for transportation, lifting and storage tend to often be high relative to 24

43 economic value at the point of use. Therefore, water at different locations may have widely differing values. Also, because of the variations in demand over the seasons of the year, the value may change with time In many places, water has little value for irrigation in winter, but it may be quite useful at that time for power generation or industry. These points are important particularly for comparing values among uses. Water in its raw (untreated) form in a river-or even in a reservoir or canal-is a distinctly different commodity than treated water delivered under pressure to a business or residence, and comparisons of value in alternative uses must recognize that point. Comparing values among uses is best performed with the comparisons made in terms of raw water supplies at some specified point of diversion Appropriate Allocation Variable: Depletion versus Withdrawal To assign an economic value to water, one must express it as a monetary value per unit water volume or quantity used. For offstream uses, the quantity variable can be either the amount withdrawn fiom the river (a gross measure) or the amount depleted or consumed (in the sense of transpiration or evaporation to the atmosphere). Values will differ greatly, depending upon which measure is chosen. The choice of withdrawal or depletion will depend on the purposes at hand. For the economist interested in predicting user behavior in response to changing prices or entitlements, the withdrawal measure is often more appropriate, because that is the measure upon which water users base their allocation decisions. Where necessary, adjustments to express benefits per unit depleted can be subsequently made. Turning to nonconsumptive or instream uses, neither of the above variables are precisely relevant. One must take any change in form, timing or location as a measure of use. In the case of evaluating instream versus offstream uses, incorporating a hydrologic model which can adjust for all these interdependent factors becomes an important aid (e.g. Booker and Young, 1994, Booker, 1995) 2.10 UNCERTAINTY AND SENSITIVITY ANALYSIS Estimating benefits for long run water investment or allocation decisions by its nature requires forecasting the behavior of a number of economic, technological and social variables for a many-year planning period. Because of the unpredictabilities in these factors influencing water management decisions, no analyst can expect to be accurate in such a situation. It is desirable that some recognition to uncertainty be incorporated into benefit analysis. Basing a plan simply on best-guess projections may bring about an unwarranted degree of confidence in the results. A number of formal treatments of uncertainty are applicable to evaluation of water investment and allocation decisions (e g. Morgan and Henrion, 1990). However, the 25

44 techniques recommended in these sources - usually based on estimating objective or subjective probabilities of occurrence of key variables- will usually require too much in the way of analytic expertise and study resources to be useful under most actual planning conditions. A more practical alternative for acknowledging uncertainty is to use "sensitivity analysis". The effect of (sensitivity to) important variables on the estimated value of water is determined by varying one element at a time to determine the sensitivity to erroneous forecasts (Gittinger, 1982). For example, an study of the economic benefits of irrigation should test for sensitivity to assumptions about crop yields and crop prices. The purpose of a sensitivity analysis is not to reduce the risk of a plan. The analysis does not, initially at least, change the facts. Sensitivity analysis show the impacts of incorrect assumptions regarding key parameters. A variation on sensitivity analysis is the "switching value" test. The switching value test investigates how far a key element in the analysis would need to change in an unfavorable direction before benefits fell below zero. 26

45 3. TECHNIQUES FOR MEASURING THE VALUE OF WATER 3.1 PRELIMINARIES For applied benefit-cost analysis (BCA), benefits and costs are expressed in monetary terms by applying the appropriate prices or shadow prices to each physical unit of input and product. Three general types of estimates are employed. First, an important source of the prices used for BCA are the result of observing market activities. So long as the observed markets are properly functioning (reflecting numerous buyers and sellers and the absence of monopoly powers and external costs), the use of market prices is appropriate. Second, when markets exist, but are not competitive, the prices observed are not proper reflections of underlying preferences or costs and need to be adjusted (for example, when agricultural commodity prices are controlled by government regulation or when minimum wage rates are set above market clearing prices). And third, in many cases it will be necessary to estimate prices for impacts which are not registered in any market (such as the value of water used for power generation, environmental preservation or recreation, or the damages from degraded water quality). The latter two cases are most typical in cost-benefit analysis for water resource planning; in these instances the nonobserved prices employed are called accounting prices or shadow prices. As discussed in the previous chapter, in applied BCA, the terms benefit and cost take on a narrow technical economic meaning. The prices used in BCA are interpreted as expressions of willingness to pay (WTP) for a particular good or service by individual consumers, producers or units of government. Willingness to pay, in turn, is reflected in the inverse demand function, which expresses the producers' or consumers' willingness to pay as a function of quantity of the good or service. The definition of value or benefits as WTP is straightforward when market prices can be observed, since the equilibrium market price represents the willingness of potential buyers of the good or service to pay-at the margin. But also for non-marketed goods, WTP is the theoretical basis on which shadow prices should be calculated. The assertion that WTP should be the measure of value or cost follows from the welfare economics postulate that public policy should be based on the aggregation of individual preferences for beneficial effects and against adverse effects. WTP represents the total value of an increment of project output, i.e. the demand for that output. WTP is the individual's best offer to purchase the increment. Therefore, benefits are defined as any positive effect, material or otherwise, for which identifiable impacted parties are willing to pay. A particularly comprehensive, rigorous and useful discussion of theory and practice of valuing of environmental resources are found in Freeman (1993). See also Anderson and Bishop (1986); Bentcover, et al. (1986); Johannson (1987); Braden and Kolstad (1991) and Pearce and Warford (1993). Economists have developed several different techniques for valuing water-related non-marketed economic benefits. These techniques vary depending upon whether the 27

46 benefits are to producers or consumers. The analysis begins with the case for which transactions can be observed, and then, in turn addresses producers' goods and consumers' goods. 3.2 OBSERVATIONS OF TRANSACTIONS IN WATER AND RELATED GOODS Values from Rentals and Sales of Water Rights Willing buyer-willing seller transactions involving exchange of money for water are uncommon However, transactions are increasingly observed, particularly in the western U.S., where growing urban demands are being met by market transfers of water from agricultural uses. For several states in the southwestern U.S., Saliba and Bush (1987) examined a series of cases, describing how water market institutions function and identifying trends in prices. Easter and Hearne (1995) and Rosegrant and Binswanger (1994) describe examples of water rights markets elsewhere in the world. Rental markets, both officially recognized and informal, are not unusual (Easter and Hearne, 1995). Rental markets serve to exchange rights for a limited period, ranging from one irrigation cycle to periods up to a season (e.g. Anderson, 1961). The prices observed are short run (comparable to payment for renting a house rather than buying it). Rental rates must be used with caution in long run planning contexts, because rental rates may be set by factors in addition to the marginal value of water. From a survey of irrigators in northeastern Colorado, Bash and Young (1994) confirmed Anderson's (1961) earlier observation that the irrigation water rental price tended to not reflect the marginal value of water. Rather, local custom in most cases requires the purchaser to cover only the seller's already sunk out-of-pocket expenses-that is, the sellers' appropriate share of the water supply agency assessments for operating costs for that year. (Casual observation in the region suggests, however, that when water is scarce due to drought, rental prices do rise above the level of assessment). Observations of prices on markets for perpetual water rights are more appropriate for estimating values in long run planning contexts, but some caution is required here, as well. The observed price for a perpetual water right is for a capital asset. However, the value often needed and conventionally used in planning and policy analysis is for an annual quantity. The analyst must convert the water right price to the annual value by use of capitalization formulas, which further requires selection of the appropriate planning period and interest rate. (See Appendix to chapter 4 for discussion of the appropriate interest rate). Most analysts follow a convention often used by the real property appraisal profession, and assume a constant interest rate, a constant annual value and a very long planning period (approaching infinity). Under such assumptions, the present value formula reduces to the expression in Eq. (3.1 ). V= Air (3.1) 28

47 where V is the present value of a stream of equal annual values A capitalized at interest rate r. If V is the observed market price of a water right, and r is known, the expression is solved for the unknown: A. The choice of interest rate is key in this derivation; the private market rate of interest which determines V may be influenced by expectations of inflation and risk that are not necessarily appropriate for social analysis. Assuming that the annual value (A) is constant over a long planning period may also be problematical. The private market value of water may in fact reflect expectations of increasing values of A, due to broad inflationary forces or to increasing real values of water. Michelsen (1994) describes historical variations in real water right prices for one of the best-known water rights markets- that operating in the Northern Colorado Water Conservancy District. Another problem with using markets for water rights in social benefit-cost analysis is that the markets might be distorted by public intervention, such as from agricultural policies designed to influence commodity prices. When, as has been the case in the U.S. and many other developed countries, crop prices are supported by government programs, water right prices would overstate the social value of irrigation water Valuing Water as a Part of a Bundle of Marketed Characteristics: The Hedonic Method An additional source of data for inferring water values are observations on real property transactions which include rights or access to water supply or quality as part of the bundle of property attributes being sold. This technique, called hedonic valuation (Palmquist, 1991) is an example of revealed preference approaches. The hedonic pricing model applies to another class of situations in which markets can provide data which can be used to measure willingness to pay for water supply or environmental quality differences. The hedonic model applies to markets for goods which have several attributes which are recognized by purchasers, but the attributes cannot be unbundled when purchasing the good. Examples of characteristics identifiable by market participants are the health or safety risks of jobs, horsepower of an automobile engine, the number of rooms in a house, or the quality of environment surrounding a recreational home. This method rests on the assumption that the price of some marketed good is a function of its different characteristics, and an implicit price exists for each of the characteristics. From a sample of closely similar marketed goods, an implicit price exists which reflects the value of the different characteristics of that good. The contribution of alternative characteristics can then be identified statistically. Freeman (1993) provides a very complete and detailed review of hedonic methods. See also Palmquist (1991). McConnell (1993) reviews the use of the approach for measuring hazardous waste damages. In natural resource and environmental economics, the hedonic pricing method has been most frequently applied to the residential housing market, for analysis of real property (land) sales price data exhibiting differing but measurable environmental characteristics (e.g. varying water supplies or water qualities). (The model has been also adapted to a number of other markets, including purchases of durable goods and services, such as automobiles, and willingness to work in risky or hazardous employment opportunities). 29

48 The housing market application hypothesizes that consumption of housing services depends on the structural characteristics of the dwelling (square footage, age, number of rooms, size of lot), a vector of neighborhood characteristics (crime risks; accessibility to jobs, shopping and parks) and location-specific environmental amenities (air or water quality). Application of the hedonic model usually posits that market outcomes can be approximated by an equation such as (3.2): pi = g(s 1, Ni, Qi ) + g, (3.2) where: pi is the sales price of the ih marketed item Si is a vector of the structural characteristics of the ith dwelling, Ni, a vector of neighborhood characteristics; Qi represents the environmental attribute(s) of interest; g is a function to be estimated with best-fit (regression) methods; and j is an error term. Under certain conditions, the function g( ) could be linear, but the nonlinear case is more likely. The partial derivative of the hedonic price function with respect to the characteristic of interest (e g/jq) yields a measure of the marginal value of the characteristic q. As an example of the hedonic approach applied to water resources, consider the case studied by D'Arge and Shogren (1988). A pair of neighboring lakes in Iowa, which are popular for water-based recreation, exhibited sharply differing water qualities. Sales prices for recreational homes on the lake with better quality water were higher, controlling for other factors, than prices of homes on the other lake. The authors regressed observed house prices near both lakes against various attributes of the sample, including area under roof, date of construction, quality of construction, size of lot, proximity to a lake, and so on. A dummy (binary) variable for the site with higher quality water provides a measure of the willingness to pay for improved water quality. The hedonic technique has to date received relatively little application to measuring values of water or water quality. (See the specific water uses discussed in chapters 4 and 5 for citations). Estimation of economic values of environmental resources with hedonic methods is quite difficult in practice, and the technique is subject to serious limitations. Although experience with real property markets shows that relatively strong conclusions can be reached regarding the value of structural attributes of the property itself, such as year of construction, size of house, size of lot, numbers of rooms, area under roof, and quality of construction. However, the value of environmental attributes-whose nature, future status and impacts may be imperfectly perceived by market participants-are more difficult to isolate. Data requirements are quite exacting. A large enough sample of transactions may be difficult to obtain. If water resources are already in public ownership, market transactions may not be available. Buyers and sellers must be able to recognize the actual 30

49 physical differences in the level of characteristics to be valued, which may be difficult when water supply and quality are highly variable. It is likely that not a few attempts at hedonic valuation have been unsuccessful in generating statistically reliable and economically plausible results, and have not been reported in the peer-reviewed literature. (The author is personally aware of two such cases). Finally, for estimating benefits of both environmental and recreational improvements, property values may capture only a part of the total benefits. Gains which would also likely accrue to other than property owners-i.e. to day visitors who travel to the area from elsewhere- are not captured by the hedonic technique Demand Functions from Water Utility Sales Data A frequently observable transaction concerning water is that occurring when a publicly owned or regulated water utility supplies water to numerous individual water users. The conditions for a free market are not met, because the buyer is faced with a take-it or leave-it price schedule from a single monopoly supplier. But because the buyer can usually take all the quantity desired at that price schedule, inferences on willingness to pay and demand can be derived if a sufficient number observations on transactions are available, and the transactions exhibit variation in real price. (See, e.g. Howe, 1983; Griffin and Chang, 1992). Household water demand, as with all water demand, tends to be very site-specific, influenced by a range of natural and socioeconomic factors. The demand relationship is represented graphically by the familiar demand curve, or algebraically as: Q= Qw(Pw, Pa, P; Y; Z) (3.3) where Q, refers to the individual's level of consumption of water in a specified time period; PR refers to the price of water; Pa denotes the price of an alternative water source; P refers to an average price index representing all other goods and services; Y is the consumer's income, and Z is a vector representing other factors, such as climate and consumer preferences. (Munasinghe, 1992). Typically the data used in such approaches are average quantities used at scheduled prices. The observations may be cross-section (data taken for the same time period) from a sample of water service agencies which exhibit a range of water rates and often variation in other factors influencing demand, such as income, rainfall, temperature. Less frequently, a time series (observations on the same entity over a period of years) is studied. Analysts would usually prefer that the data come from observations on water use behavior of individual households, (e.g. Nieswiadomy, 1993; Hewitt and Hanemann, 1995) but the extra expense is often prohibitive, so the most frequently seen format employs aggregate data such that the consumption and price are the average of one sup- 31

50 plier for one time period. The general approach is to apply statistical regression analysis to estimate the parameters of a demand equation. We discuss this approach in more detail in the Chapter 5 under the subject of residential water demand. 3.3 TECHNIQUES FOR VALUING WATER AS AN INTIERMEDIATE GOOD The most extensive uses of water are for intermediate or producers' goods, that is for production of goods that are not final products. Water used for crop irrigation, in industry, or for much of hydroelectric or thermal power generation are examples of intermediate good use uses For intermediate goods, the theory of producer's demand for inputs provides the conceptual basis for valuing the economic welfare implications of increments or decrements to input supplies (Just, Hueth and Schmitz. 1982, Freeman and Harrington, 1990, Freeman, 1993; Johannson, 1993) A model for producers' demands for water can be obtained from the theory of a cost-minimizing producer QW= Q,(P,, Pi, PanI XI S) (3 4) where, as before, P, and Pa represent the price of water from the given system and from an alternative source; Pi represents a vector of prices of inputs (capital, labor and materials); X stands for the quantity of product to be produced and S represents a vector of other factors, such as technology and climate The producers' demand for an input is its value of marginal product (VMP); hence, the measurement problem requires an approximation to VNIP. Three methods of valuing water as an intermediate good are in common use; the residual approach, the hedonic (land value) approach and the alternative cost approach The Residual Imputation Approach The "residual" method is the most frequently-used approach to applied shadow pricing of producers' goods, particularly for irrigation water In the residual method, the incremental contribution of each input in the production process is determined. If appropriate prices can be assigned-presumably by market forces-to all inputs but one, the remainder of total value of product is imputed to the remaining or residual input (Heady, 1952). 32

51 Derivation of the Residual Value The derivation requires two principle postulates. First, competitive equilibrium requires that the prices of all resources are equated to returns at the margin, as explained in more detail below. Profit-maximizing producers are assumed to add productive inputs up until the point that value marginal products are equal to opportunity costs of the inputs. The second condition requires that the total value of product can be divided into shares, so that each resource is paid according to its marginal productivity and the total value of product is thereby completely exhausted. Euler's theorem shows that the Total Value Product will be exactly exhausted by the distributive shares only in the event that the total value function is homogeneous in the first degree (Heady, 1952, pp ; Debertin, 1986). To illustrate: consider an agricultural production process in which a single product denoted Y is produced by four factors of production: capital (K), labor (L), other natural resources, such as land (R) and irrigation water (W). The production function is written: Y = f(k, L, R, W) (3.5) If competitive factor and product markets can be assumed, prices may be treated as constants. We may write, by the second postulate: TVPy = (VMPk X Qk) + (VMPL X QL ) + (VMPR X QR) + (VMPW X QW) (3.6) where TVP represents Total Value of Product Y; VMP represents value marginal product of resource i, and Q is the quantity of resource i. The first postulate, (which asserts that Pi = VMPi) permits substituting Pi into (3.6) and rearranging: TVPY - [(Pk X Qk ) + (PL X QL ) + (PR X QR )]= PW X Qw (3.7) On the assumption that all variables in (3.7) are known except Pw, that expression can be solved for that unknown to impute the value (shadow price) of the residual claimant, (water) Pw*, as follows: Pw= {TVPy- [(PK X QK ) + (PL X QL) + (PR X QR )]}/Qw (3.8) 33

52 The "Change in Net Income" and other Variants of the Residual Approach The pure case of residual imputation (for a single-product firm) is only of interest as a special case. The more typical problem will be to estimate values for a multiproduct operation. Often, too, the task is to estimate the value contributed by a partial increment or decrement to the water supply, rather than that for the entire quantity. A method developed for such situations is called the "change in net income" (CINI) approach, to which we now turn. (This method is a form of what is sometimes called "the Valuation of Productivity Change" or VPC approach to intermediate goods). The willingness to pay for an increment of water is the net producer income associated with that increment. This approach can be represented by a process very similar to that used for residual imputation. [The U.S. Water Resources Council (1983) recommends a process which is essentially the CINI method for irrigation benefit evaluation.] A more general multi-product/multi-input production function can be written as: f(yi,. Ym, XI.... Xn) 0 (3.9) where Y is a vector of outputs of feasible products and X is a vector of production inputs. The net income (denoted Z) from producing a given set of products can be represented by: Z = zi=1 m(yi* x Pyi) - Xj.l...n (Xj* x Pxj) (3.10) The Change in Net Income is: AZ = Z, - Zo (3.11) where the subscripts 0 and 1 refer to the "without project" and "with project" situations, respectively. The Change in Net Income method is readily adapted to solution on a nicrocomputer with conventional spreadsheet software General use of Residual Imputation The residual technique and variants of it are extremely sensitive to small variations in assumptions about either the nature of the production function or about prices. Biases can be introduced by any of several types of errors or omissions. Therefore, the approach is most suitable for cases where the residual claimant (in this case, water) contributes a significant fraction of the value of output. If an input that should be represented in the production function is omitted, this will permit the contribution of that input to be attrib- 34

53 uted to the residual claimant, thereby overstating the value of water. A similar but lesser bias arises if the amount of input, such as labor, is merely underestimated rather than omitted entirely. The assumptions of residual imputation are not overly restrictive, but care is required to assure that the conditions of production under study are reasonable approximations of the conceptual model. The main issues can be divided into two types: those relating to specification of the production function; and those relating to the market and policy environment (i.e. the pricing of outputs and nonresidual inputs). We address each of these in turn Conceptual Problems Arising with Residual Intputation: Specifying the Production Function Two types of problems may be identified when attempting to accurately specify the production function. One has to do with listing of all relevant inputs (other than the residual) and assigning productivities to them. If one or more important factor inputs are omitted from the specification of the production function, the productivity of the omitted input(s) is erroneously attributed to the residual claimant. The value productivity of the residual would be correspondingly overstated. For example, in the case of evaluating benefits of crop irrigation, if the costs of labor are omitted from the calculations, the contribution of labor is implicitly added to the residual attributed to water. The omission of variables most often occurs when a short run modeling framework is used to represent a long run planning situation. In short run contexts, it is appropriate to treat certain production inputs as fixed, and to omit charges for those fixed inputs in calculating the residual value. For example, Michelsen and Young, (1994) employed a short run concept of benefits foregone to estimate the price at which farmers would be expected to be willing to sell a an option to rent their irrigation water during a drought. Although short run models have their place, the long run is appropriate in most water development and allocation planning contexts, and in those cases all factor inputs should be treated as variable and costs assigned to them. A second issue which occurs in specifying the production function is accurately forecasting the levels of output associated with given factor inputs. Over- or underestimates of the level of production possible from a given bundle of inputs will bring about corresponding over- or underestimate of the residual value. Because of the arithmetic of the process, any degree of error in forecasted revenue from incorrect output forecasts will be magnified several times in the residual. The mutual interdependence of technological advances, levels of investments and product prices must be emphasized. Harberger (1974, p. 19) made the general point over two decades ago: Almost any investment made today would become profitable if no competing investments were made in the future... The "profitability" of today's in- 35

54 vestments should be estimated on the assumption that all "profitable" future investments will also be made...here, of necessity, the project analyst himself has to estimate an expected time path of prices-not on the assumption that his project will stand alone, nor on the assumption that future projects will be held up in order to "protect" his current project, but on the much more rigorous assumption that future investments will be made on their own merits. A third concern relating to specifying the role of water in a production function arises with difficulties in empirical measurement. This problem is found in cases in which water contributes a relatively minor portion of the total value of output, in which event, potential errors in assigning opportunity costs to nonwater inputs may cause a serious error in the residual assigned to water. For example, in valuing water used in industrial production, analysts employing the residual technique must assign opportunity costs to other inputs, such as financial capital. A small change in the assumed opportunity cost of stockholder financial capital or in the interest rate paid on debt can have a large impact on the residual derived for water Assigning Prices to Inputs and Outputs If government intervention or market failures lead to prices for input factors and products which deviate from the competitive equilibrium prices, then the imputed value of the residual will accordingly be inaccurate. Incorrect prices obviously translate directly into biased estimates of the value of the residual claimant with opposite sign. Underpriced inputs yield overestimates of the residual, and the converse. Turning to outputs, if they are underpriced, the residual is underestimated. The converse is again the case. These are frequently encountered problems in water resource planning: governments frequently intervene in the workings of markets to raise or lower prices of either products or inputs. Some countries, for example, have kept agricultural commodity prices below world market prices with policies designed to keep food prices low to urban consumers. When artificially depressed prices are lower than market equilibrium, then the imputed values of irrigation water are accordingly lower than they would other wise be. Conversely, in developed countries such as the United States, national agricultural policies raise prices for a number of commodities (e.g. rice, wheat, maize) above what would be under an unregulated market condition. In this case, the residual value of water is overstated from the national economic perspective. On the input side, wage rates may be raised by minimum wage regulations. Capital costs are affected by government monetary and fiscal policies influencing interest rates. Electric power prices are often found to be less than the marginal cost of producing additional power. An understatement (overstatement) of input costs yields a higher (lower) residual value than should be the case. 36

55 Residual Imputation in Long-run versus Short-run Planning Contexts The immediately previous discussion implicitly assumed a long run planning context, one in which all resources are considered as variable. In some cases, the analysis may be for a short run situation. In such event, some assets may be assumed to fixed and therefore these fixed costs are ignored for the decision at hand. For example, in agricultural crop irrigation, evaluations of annual cropping decisions in the face of temporary water shortages may be made in terms of maximizing the return to all fixed factors. In such cases a short run value, which would be higher than a long run value in the same situation, would be entirely appropriate. There are a number of potential variations in calculating a residual- ranging from the very short run to the long run. The message is that caution must be exercised to assure that the appropriate concept is being used for whatever situation is being analyzed Mathematical Programming (Optimization) Models as an Application of the Residual Imputation Technique Mathematical programming (MP) models are often used to value water as a producers' good in agriculture and industry. Such models often are developed to represent the optimum allocation of water and other resources so as to maximize profits, subject to constraints on resource availability and perhaps to institutional constraints (Williams, 1990). The objective function for a mathematical programming model may be written: Max f(7t, X), (3.12) subject to a set of constraints: A'X < B where mri represents net return per activity; Xi are production activities, the elements of the A matrix are production coefficients and B is a vector of constraints on production inputs such as labor, capital, natural resources, etc. (Williams, 1990). The objective function is usually linear, but can be nonlinear (i.e. quadratic, etc.). In interpretations of mathematical programming models for purposes of water allocation, the parameter ni is usually a measure of the marginal return to water in activity Xi. Mathematical programming is advantageous in situations where a wide range of technological options (formulated as alternative activities) is to be studied; the model can find the profit maximizing set of activities (X ) given the constraints on resources. Solutions of 37

56 a MP model for a range of water supply constraints trace out a set of net total benefit points, from which a set of net marginal benefit points can be then derived- It is important in developing the model that the marginal value productivity represented by the net benefit coefficient 7ri is accurately calculated according to the model of residual imputation described previously. Costs of all nonresidual inputs should be deducted as appropriate. Analysts sometimes seem to give more attention to the theoretical formulation of the model and its solution than to the calculation of appropriate parameters. 3.4 ERRONEOUS WATER VALUATION WITH A VERSION OF THE RESIDUAL TECH- NIQUE: THE "VALUE ADDED" APPROACH FROM INPUT-OUTPUT MODELS. Applied methods of valuing water in withdrawal and intermediate good uses have not received the intensive critical scrutiny which has been devoted to determining environmental and nonuse values. Therefore. some less-frequently used approaches which tend to overestimate actual willingness to pay for intermediate good values continue to appear. One particular approach to valuing water in production, from the value-added calculations made in regional interindustry (also called Leontief Input-Output) analysis will, unless extra care is taken, overstate the correct economic value of water as an intermediate good. The value-added approach appears, at first glance, to be similar to the residual method described earlier. However, it differs in certain key respects. The critique outlined below is developed more frilly in an earlier paper (Young and Gray, 1985) The Input-Output (Interindustry) Model Applied to Water Valuation The input-output (1-0) model is a static model of production, usually used to model a geographic region or political subdivision for purposes of understanding the structure of a regional economy, and for making short-run predictions of the effects of exogenous changes in final demands on such variables as output, employment and income. 1-0 models can predict such variables not only in the aggregate, but for each sector of an economy. The conventional 1-0 model is characterized by a production function exhibiting constant returns to scale. The model lacks constraints on resources, contains no specified production function and does not allow for choice on either the production or consumption side. Factor and product prices are constants in the conventional 1-0 model. A conventional input-output model representing a region (usually for time period of a year) views the economy from two interrelated perspectives (Miller and Blair, 1985). The first viewpoint is according to the five ways that the annual outputs of its industries are distributed: to household consumption; to investment in capital goods; to consumption by government agencies; to inputs used by other individual industries or to exports beyond the regional economy. The other perspective-which is our primary interest here- is according to payments for inputs: to regional suppliers of inputs; to wages and salaries for 38

57 the work force; to rents paid to land and other natural resources; to interest and depreciation on capital ; to profits; and to any imported from outside the region. This latter formulation can be expressed (for industry i) in terms of the Total Outlay (X) of industry i: Xi xi..nxj + (Wi + Li + K, + Pi + Si) + Mi (3.13) where: xij =the amount of industry i's outputs used as inputs by industry j Wi = wages and salaries paid to workers in industry i L= rents paid to land and other natural resources used in industry i Ki = interest and depreciation on capital used in industry i Pi = profits paid to capital owners in industry i Si taxes paid to government by industry i Mi = paid for inputs imported from outside the region Consider the terms inside the parentheses in Eq. (3.13): (W; Li, K, ; Pi; Si). These, in total, constitute what is termed the value added (also termed Gross Regional Income), representing the payments to factor owners in the region. Denote the value added as VA. The value-added method of imputing a value to water in a specific sector can be represented as follows, employing a sectoral production function similar to that shown in Eq above. The sector will represent an aggregate of a particular industry, such as manufacturing, food processing, etc. In the VA imputation procedure as usually performed, imputation of the "value productivity of water" P,* is accomplished by rearranging (3.10): VA = Xi - 1j=l... Xij - Mi = (Wi + Li + Ki + Pi + Si) (3.14) and dividing VA, by the quantity of water (W 1 ) used in sector i: P,* = VAi/Wi (3.15) Interpreting the "Value of Water" Derived from Value Added This imputation process shown in 3.11 and 3.12, at first glance, appears to be equivalent to the residual approach discussed above. However, on reflection, it will be recognized that rather than isolating only the contribution of the water resource, the proc- 39

58 ess imputes the productivity of all primary resources (labor, capital, depreciation, other natural resources, etc.) to the value of water Clearly, dividing a sector's value added by its water use yields a figure which greatly overstates the theoretically correct residual value of water. To understand the bias in the approach, note that the opportunity costs of all of these other primary resources are implicitly assigned a zero shadow price. Although there are cases where some primary resources might be shadow priced at lower than market price, to assume that the shadow price of all primary resources is zero and thereby attributing all primary resource productivity to water clearly yields a large overstatement of its correct value Another way of expressing this critique is to note that certain payments to capital and labor are treated incorrectly as income or benefits (rather than opportunity costs) in the value added approach. In any case, evaluating regional development projects on the basis of a value-added concept would greatly inflate the estimated returns to a public investment program in comparison to gains possible from private use of funds or from competing public investments. The important point of this discussion is. although the terms seem to mean the same thing, value-added derived from an input-output model is not a measure of marginal value productivity. It is the latter concept that we wish to measure in deriving conceptually proper estimates of the value of water. To conclude: the "value added" method of imputing a marginal value productivity or opportunity cost to water resources used in industry or in agriculture will not yield a measure commensurate with other measures of water value, and will greatly overstate the "correct" value for economic analysis. This is not to say that input-output models cannot be a building block in the process of estimating the marginal value productivity of water. With some additional analysis, a more accurate measure can be derived. Using a value added format from an inputoutput model to estimate the value of water will require deduction of the opportunity costs of all nonwater primary inputs (e.g. labor, capital, and natural resources) from value added to obtain the appropriate measure of the residual return to water. 3.5 THE "ALTERNATIVE COST" APPROACH Another method appropriate to evaluating water-related intermediate good benefits is the alternative cost approach, (which can also be used for consumption goods). The technique is attractive under the assumption (valid only in certain limited instances) that if a given project of specified output costs less than the next-best public or private project which can achieve the same output, then the cost of the next best project can be assigned as the benefit to the public project under consideration. However, the analysis must verify that the higher cost alternative would actually be constructed in the absence of the project under consideration. Put another way, the effective demand must be established for the alternative product. In the field of water resource planning, the alternative cost approach has been employed for evaluation of many types of benefits, including municipal, industrial, hydroelec- 40

59 tric and thermal electric power. When estimates of a direct demand schedule proves difficult because of lack of data or other reasons, the alternative cost approach may be a solution. However, the approach is subject to particular limitations, and should be used when its applicability is assured. Steiner (1965) remains the primary authority on the alternative cost technique, and we follow his argument below. (See also Herfindahl and Kneese, 1974; pp for an exposition and extension of Steiner's analysis, illustrated with examples from inland waterways navigation and water quality). The alternative cost approach is based on the notion that maximum willingness to pay for a publicly supplied good or service is not greater than the cost of providing that good or service via some other process or technology. The alternative, which may be produced from either the public or private sector, should be a substantially different means of producing the same output. In addition to being used as a measure of benefit or willingness to pay under certain limited circumstances, the alternative cost concept plays an important role in the separable costs-remaining benefits method of allocating joint costs in multipurpose water projects (H.P. Young, et al, 1982; H.P. Young, 1985). The alternative cost technique is easily misused, and should be applied only with caution. The main limitation is that some alternative could always be conceived which would be more expensive than the project being evaluated, thereby inevitably producing an estimate of positive net benefits. Therefore, the alternative cost method must be supplemented by a study to confirm that the demand for the alternative is sufficient to justify the alternative expenditure. Performing the alternative cost test is not difficult in concept, although accurately developing the detailed empirical analysis may require considerable time and effort. The alternative cost test is but another application of discounted cash flow analysis. The present value of costs of each alternative are calculated on the basis of commensurate planning period, price level, discount rate and the like Questions quite similar to those arising with the residual technique, must be addressed. These include: selecting the long run or short run context, and carefully forecasting trends in technology and prices over the planning period for the alternative investment. Some analysts have dismissed the alternative cost analysis as merely a form of cost-effectiveness study which should be performed in the course of any competent economic evaluation, and doesn't warrant special treatment in a monograph such as this. That each proposed plan should be tested to assure that it is the least-cost alternative is of course correct, but the alternative cost technique plays an additional role in measuring benefits of water-related projects and programs. As will be seen in the next chapters, the alternative cost approach provides a tool for estimating a price for certain unpriced benefits, such that when combined with residual imputations calculations, is valuable for imputing benefits of water use in hydroelectric power and waste load dilution. 41

60 3.6 MEASURING WILLINGNESS TO PAY FOR WATER AS A CONSUMPTION GOOD: OVERVIEW Turning next to estimating the consumers' willingness to pay for water, we address the two types of goods: those that are private goods (rival in consumption) and public goods (nonrival). Demands for water as a private consumption good are found primarily among the residential water users, pastly treated in Section above. Public good demands are for recreation, fish and wildlife habitat and amenity values. There will, however, be some instances where it is difficult to draw the boundaries between these categories User Surveys to Determine Willingness to Pay for Water-Related Public Benefits In those cases where water yields a public or collective good, neither diversion for production nor purchase prices exist, and special data collection and demand evaluation methods must be adopted. These cases often are associated with recreation or aesthetic enjoyment of water in its natural surroundings or with water quality improvement. Several different approaches to measuring benefits and costs of instream and public benefits are commonly employed Freeman (1993) provides perhaps the most authoritative and exhaustive treatment to date. Pearce and Turner (1990), Dixon, et al (1994) and the individual chapters by specialists in Braden and Kolstad (1991) also offer useful recent reviews of the issues. See also the somewhat earlier presentations by Anderson and Bishop (1986), Bentkover, et al, (1986) Johansson, (1987) and Randall (1987). Two general lines of approach have been developed, both based on user surveys of actual or hypothetical behavior (Freeman, 1993). One broad approach is termed the observed indirect or sometimes the revealed preference method. Relying on observations of actual expenditure choices made by consumers (revealing their preferences) the method infers net willingness to pay from the differences in expenditures observed with varying levels of environmental amenities. If the use of water-based recreational services influences the demand for any marketed commodity, observations on purchasing behavior related to the marketed commodity can be analyzed to derive information on the preferences and willingness to pay for the environmental amenity. A second line of approach is represented by what are often termed the "hypothetical questioning" methods. They involve asking people directly about the values placed on proposed or hypothetical improvements or reductions in environmental services. 3.7 REVEALED PREFERENCE OR OBSERVED INDIRECT METHODS 42

61 3.7.1 The Travel Cost Approach The travel cost method is the most widely used example of the observed indirect methods. The well-known economist Harold Hotelling is credited with the observation (in 1949) that while the absence of variation in fees for recreational and amenity sites precludes estimating directly the demand for such sites, if the cost of travel to a recreational site varies widely among consumers, and if these consumers respond to higher travel costs in the same way that they would respond to higher entrance fees, the analyst can derive a demand schedule for recreation at the site from the costs of travel. Marion Clawson began to operationalize the proposed approach, and later, working with J. Knetsch (see Clawson and Knetsch, 1966) continued to refine it The underlying assumption of the travel cost approach is that observable recreationist behavior as related to increasing costs of travel reflect the changes in demand for the activity which would occur if prices were actually charged (See Walsh, 1986 for an introductory discussion. McConnell, 1985; Freeman, 1993; Fletcher, Adamowicz and Graham-Tomasi, 1990; and Ward and Loomis, 1986, provide more formal developments and critical evaluations). The travel cost approach involves two steps: the first is to estimate the individual recreationist's demand for the resource, and the second is to statistically derive the relevant aggregate resource demand curve. In the most basic approach, concentric circles of different radii are drawn around a particular site. A sample of recreationists must be contacted and questioned regarding the number of visits, distance traveled to the site and their actual travel expenditures. The data collection procedure is designed to question respondents from a range of travel distances, whose expenditures will vary with distances traveled. Consumer surplus can be calculated for each zone by finding the area below the demand curve and above the cost of travei for residents of each zone Note that the costs of travel themselves are not a measure of the site value, those costs are used only to infer the desired consumer surplus measure. The principal attraction of the travel cost approach is that like other revealed preference approaches, it reflects actual consumer choice behavior. Most economists agree that when possible, this is preferable to methods which rely on responses to questions regarding hypothetical scenarios. However, a number of problems are encountered in application of the travel cost method. The approach has mainly been applied to estimating consumer surplus for access to a specific recreational site, to determine benefits associated with developing or maintaining such a site. For present purposes, a main concern is that water is likely to be only one of many attractive attribute of a site, and people travel to rivers or lakes for a multiplicity of reasons, some of which may be unrelated to water supply or quality. To obtain the value of the water or of a water quality improvement, some method must be devised to isolate the contribution of water to the total estimated site value. One solution to this difficulty is to perform a multiple site analysis. If the sites vary according to, say, water quality, it is possible to also infer the incremental value of the improved quality from a travel cost analysis. In a study notable for its conceptual and statistical rigor and for its careful documentation of procedures and results, Smith and Desvouges (1986) developed estimates of the value of improved water quality on a sample of U.S. Army Corps of 43

62 Engineers reservoirs. (They also employed, for comparative purposes, the contingent valuation method to the same topic.) Smith and Desvouges developed a generalized travel cost model designed to infer the value placed on water quality improvements by recreationists. However, the generalized model yielded results that were implausibly large, illustrating the problems of extrapolating beyond the range of the site characteristics for the model was estimated. The study is also noteworthy for the large amount of resources-both effort and money-which must have gone in to the effort extending over a five-year period. Several important judgments must be made in any application of the travel cost method. The calculation of travel costs is not at all straightforward. One problem is specifying the out-of-pocket costs of travel. For example, should the cost of the automobile reflect only the variable costs of the trip (gas, oil, etc.) or should some estimate of wear and tear on tires and auto be included? If food and lodging costs are incurred, should the actual costs be assessed-which might reflect payment for a discretionary experience unrelated to the outdoor recreation site, such as a special restaurant meal or a nice hotel- or should a minimum alternative cost be recorded? Different analysts have made differing assumptions on these matters. It can be easily seen that travel cost estimates are quite sensitive to the particular judgments adopted. Another significant issue in estimating costs of travel is how to account for the opportunity cost of recreationists' time. As with the out of pocket costs, the higher the estimated opportunity cost of time, the higher the resulting derived value of the recreational experience (Freeman, 1993:448-54). What portion of time are to counted as recreational activity, and how should time be priced? Some have argued that time should be valued at the wage rate, although others suggest that if work has a disutility, the opportunity cost of leisure is less than the wage rate. In addition to considerations of costs and travel time, a number of other problems have been identified with the travel cost method. Among the more important of these is the treatment of substitute sites. As with any economic good, the availability and cost of substitutes are significant determinants of demands for recreational sites. If relevant substitute recreational activities are not accounted for in the analysis, the estimates of consumer surplus will be biased. In a study designed to evaluate the recreational benefits of a proposed Corps of Engineers reservoir, Burt and Brewer (1971) first formulated a multiple site travel cost model to account for substitute reservoir sites. Other issues include the appropriate functional form from which to derive consumer surplus measures, problems of aggregation from a sample to a population, and how to treat multiple destination trips (For discussion of these issues, see Freeman, 1993; Fletcher, et al, 1990; McConnell, 1985). Although the travel cost model has the advantage of measuring revealed preferences, its use remains problematic for many applications in water resource valuation. Its applicability in developing countries is likely to be quite limited. In addition to the remaining unresolved matters of judgments in its application, to apply the technique to accurately to estimate the contribution of site characteristics requires a very large study budget. 44

63 3.7.2 The Hedonic Price Approach The hedonic pricing model applies to another class of situations in which markets can provide data which can be used to measure willingness to pay by consumers for water supply or environmental quality differences. The hedonic model applies to markets for goods which have several attributes which are recognized by purchasers, but the attributes cannot be unbundled when purchasing the good. Water quality represents such case. (See the discussion of the hedonic pricing model above in Section 3.2.2). 3.8 QUESTIONING CONSUMERS REGARDING VALUATION OF HYPOTHETICAL ENVIRONMENTAL CHANGES-THE CONTINGENT VALUATION APPROACH Many important economic evaluation problems occur for which no value measures can be derived from observing individual choices through a market. For example, because of the public good aspect of water quality and environmental management issues, water and environmental resource managers in public agencies often encounter such cases. Other examples where actual consumer choices are nonobservable are cases where the policy change is potential rather than actual. Freeman (1993) calls the methods developed to measure environmental values in such cases "hypothetical methods". Carson, (1991) offers the term "constructed markets". In this approach, respondents are offered conditions simulating a hypothetical market in which they are asked to express willingness to pay for existing or potential environmental conditions not registered on any market. The most common form of questioning on hypothetical futures is called the "contingent valuation method" (CVM). It involves asking people directly what they would be willing to pay contingent on some hypothetical change in the future state of the world. (See Mitchell and Carson, 1989, for an exhaustive and balanced description and evaluation of the contingent valuation technique, one that will doubtless remain a standard reference on the subject for some time to come. Cummings, Brookshire and Schultze, 1986, provided an earlier general assessment). For example, the question might appear as follows: Suppose the management of the _ River is changed so that the flow during the month of is increased by an average of m 3 per second. What is the maximum amount you would be willing to pay for this increase? Many analysts have applied the CV Method to water- related issues. A limitation for ascertaining the marginal value of water may occur because the questions asked do not relate to incremental changes in water supply or quality, but to the value of the site or policy itself However, questions regarding alternative amounts of water for fishing, 45

64 boating or streamside recreation, illustrated with photographs of alternative situations have elicited useful estimates of the marginal value of streamflow (Daubert and Young, 1981; Loomis, 1987; Ward, 1987). Several applications of CVM to the measurement of water quality-related benefits are described by Smith and Desvouges (1986). These approaches are treated in more detail in Chapter Methodological Issues in Designing a CVM Study Duffield, et al, (1994), quoting a conference paper by Bishop and Heberlein, identify six key methodological choices in devising a CVM study. The first is the target population. For studies of direct use values, the sample population will be from direct recreational water users, while nonuser values, such as existence, bequest and option values must study a regional population. Sampling procedures must be devised so that the sample represents the target population (Salant and Dillman, 1994, Fowler, 1992).The second choice is product definition. The resource flow being studied must be clearly described to respondents. Estimating values of changed flows is difficult to communicate. Daubert and Young (1981) are among those who have employed visual aids such as charts or photographs. A payment vehicle, which represents to respondents the way that the amenity would be paid for. Taxes or site fees are examples. The vehicle should be realistic and plausible way of collecting revenue, and it should also be noncontroversial as a financing method. The question format represents a major decision. The questions may be openended, bidding games or dichotomous choice forms. (These are described and discussed in more detail below). The fifth decision is the method of statistical analysis to be used. This is dependent on the question format; regression methods apply to most approaches, but dichotomous choice questions require discrete choice statistical models such as the logit approach. Sixth is the list of supplemental data to be incorporated into the demand analysis as shift variables. One decision not mentioned in the above list is the choice of data collection technique This can be personal interview, telephone interview or mailed questionnaire. There are tradeoffs in accuracy and cost that need to be decided on a case by case basis. (See, among others, Salant and Dillman, 1994, for a discussion). A CVM questionnaire typically comprises three components (Mitchell and Carson, 1989; Freeman, 1993): The first component describes the setting under which the respondent is to imagine herself or himself In the case of application to water allocation or policy issues, a description is provided of the water resource or water-based amenity to be valued and/or the conditions under which some policy change is being undertaken. Second are the choice questions which will be used to infer values of the amenity or policy change. (The alternative forms by which these questions might be expressed are discussed further below.) 46

65 The third element asks questions about the respondents. A CVM study will, as will any study of economic demand, be improved by including the usual demand shift variables, such as socioeconomic characteristics, (age, education, income and gender). Also in this category are questions soliciting information on attitudes and beliefs, such as attitudes toward environmental policies. This socioeconomic and attitude information is used shift variables in the subsequent statistical analysis Forms of Questions Used in CVM Studies The way that willingness to pay is solicited in contingent value research has evolved over the past two decades of extensive research into the method. We discuss below the five main approaches which have been applied in the CVM literature, although there are some variations on these which make the potential list longer. One of the initial approaches was to request a direct expression of value for the good in question. In one variant, the analyst posed a direct open-ended question: How much would you be willing to pay? Called the Direct Question approach, this method is now recognized to put the respondent into an unfamiliar situation. (In the type of market settings familiar to people, they are presented a listed price that they can accept or ignore.) High rates of nonresponse or relatively large numbers of implausibly high or low valuations were experienced with the direct question approach. An iterative method, called the bidding game was another form of elicitation used in many early studies. Respondents are asked if they were willing to pay (bid) a specified amount-call it B. In the event of an affirmative response, the question was repeated with a series of higher prices, until a "no" response is received. Conversely, if the initial response was to the amount B was "no", the questions iterate downward until a "yes" is recorded. With the bidding game approach, tests were made to ascertain whether the starting value (B) had any influence on the bid. The results of these tests were in the affirmative. The final bids tended to be correlated with the initial value offered to the respondent; the initial bid offers a cue as to the value to the respondent. Therefore, what has come to be called 'starting point" bias appears to be a problem with the bidding game form of questioning (Cummings, Brookshire and Schultze, 1986). Another form of the direct questioning approach is the use of a "payment card", first used by Mitchell and Carson in the early 1980s as an alternative to the bidding game. (See Eitchell and Carson, 1989: for a description). A payment card visually lists for the respondent an array of potential bids of annual willingness to pay, ranging from zero to some very large number. Mitchell and Carson augmented the list of potential annual payments by also identifying the amounts already being paid through taxes for families in their income group for other, unrelated public goods ( such as police and fire protection; national defense; roads and highways and public education). The payment card approach has been found to reduce but not eliminate the potentiality of starting point bias. 47

66 The initial application of the "take it or leave it" or discrete-choice method has been attributed to Bishop and Heberlein (1979) to a study of willingness to pay for goose hunting in Wisconsin. They established a number of predetermined prices, which are expected to include the entire likely range of willingness to pay. The respondents are grouped into subsamples, and each member of a given subsample is presented with the same one of those selected prices, and asked to give a yes or no answer to whether she or he would be willing to pay that amount for the amenity. Each price is assigned to a randomly selected subsample of respondents. This type of judgment is similar to that made by consumers in market contexts. Because the method is similar to the experience of voting in a referendum election, where the voter expresses approval or disapproval of a specific proposal, the term "referendum" is has also come into use to identify the take it or leave it method (Freeman, 1993). The approach is simpler for respondents than the bidding game, and exhibits what economists call "incentive-compatibility"-the respondent has no incentive to strategically respond in ways to bias answers toward desired policy outcomes. The logit-model statistical procedures used to analyze referendum-type (dichotomous choice) questions predict the probability of accepting an offer as a function of the stated bid and of socioeconomic variables. The econometric methods appropriate for logit models may be more complex or unfamiliar than those appropriate to direct question or bidding game responses. Dichotomous choice approaches are subject to a form of starting point bias, what Mitchell and Carson call "yea-saying", the tendency of some respondents to agree with the interviewer, no matter what their own opinions might be. Another important limitation of the referendum approaches is, because the responses do not show the respondent's maximum willingness to pay, a larger sample size is required to measure the willingness to pay function. The last contingent valuation variant mentioned here is the take-or-leave-it with follow-up (Mitchell and Carson, 1989: 103). Proposed as a way to circumvent the more demanding sample size needs of the basic referendum method, this approach calls for a second question to follow up the first. If the first response is a "no", a randomly selected second lower bid is posed; if the first response was a "yes", a randomly selected higher second question follows. Gains in the information content per response are achieved because the follow-up more often brackets the true willingness to pay Potential Sources of Systematic Error (Bias) in Contingent Valuation Studies Mitchell and Carson (1989: chapter 11) evaluate three types of potential sources of bias in CV studies. One is a scenario which provides incentives for misrepresentation of their true preferences. The questionnaire scenario can encourage strategic behaviorresponses deliberately chosen to influence future provision of the amenity being valued. Critics of the approach have been greatly concerned with the likelihood of strategic behavior, but the evidence suggests that properly designed questions can avoid or minimize this aspect, and respondents seem to make a serious effort to convey their own feelings (Smith and Desvouges, 1986). A plausible payment obligation seems to be a key to avoiding strategic bias. A second type of misrepresentation of true preferences is compliance bias-a tendency to fit their responses to the perceived preferences of the inter- 48

67 viewer or the surveying organization. Interviewer effects are an issue many types of surveys-not only with CVM- and careful training and supervision of interviewers and the use of experienced interviewers is the path to minimizing the problem (Salant and Dillman, 1994; Fowler, 1993; Rea and Parker, 1992) A second source of bias has been mentioned above: the possibility that the scenario provides implied value cues to the respondent about what an appropriate response might be. For example, starting point bias was mentioned above as a problem with the bidding game approach. Relational bias is caused when the amenity to be valued is linked to other public goods. The payment card method, which provides explicit statements of the cost of alternative public goods, was thought to be subject to relational bias. However, tests by Mitchell and Carson did not lend support to this concern. Importance and position biases arise by the way the amenity is presented in the questionnaire. The respondent should not be led by the fact of the survey itself or by the position of the willingness to pay elicitation in the questionnaire to infer an extra value to the amenity. Assuring that respondents are comfortable with zero responses, when that is their actual valuation, is accomplished by a statement indicating that as in a referendum election, some people will vote no while others vote yes. A third and last general category of potential bias can result from scenario mispecification. This situation occurs when the respondent fails to understand the scenario intended by the researcher. If the respondent does not comprehend the scenario, and the misunderstanding causes a systematic over or understatement of responses, this form of mispecification can result. Survey researchers have long been aware that small variations in wording can greatly affect responses. The antidote in this case is careful questionnaire design, including the use of focus groups and extensive pretesting of questionnaires to assure that the intended meaning is being conveyed Advantages and Disadvantages of the Contingent Valuation Method As with any technique for measuring water value, the advantages and disadvantages of CVM need to be considered as one makes the choice to use it. The principal advantage of the method is that it can potentially measure the economic benefits (or damages) of a wide assortment of beneficial (or adverse) effects in a way that is consistent with economic theory. A major plus is the possibility of evaluating proposed, in addition to already available, goods or services. The technique can be addressed to values, such as nonuser values, that cannot be successfully dealt with by any other approaches. As discussed in Chapter 5, the method has been successfully adapted to studying demand for domestic water and sanitation improvements in rural villages in developing countries (Whittington and Swarna, 1994). There is a downside, as well. Although a contingent value study can be an effective measurement tool where no other technique applies, if one hopes for an accurate result, extreme care must go into design and conduct of the survey. All the problems of sample surveys must be recognized and overcome (Salant and Dillman, 1994; Fowler, 1992). Questionnaires must be carefully formulated and tested and if not a mail survey, 49

68 interviewers carefully selected, trained and supervised. Econometric analysis of the data may present challenges. CVM studies, properly performed, require a significant research effort, well-trained staff and a budget to match. 3.9 THE "UNIT DAY VALUE" TABLE: A VALID SHORTCUT METHOD FOR MEASURING RECREATIONAL BENEFITS? One other basis for obtaining recreational benefits might be noted here. Because of the time and expense required for direct user surveys, federal agencies in the U.S. (including the U.S. Water Resources Council) have developed generalized tables which purport to assign a consumer surplus value per visitor day which can be used for project appraisal purposes. The U.S. Water Resources Council (1983, pp ) uses a measure called a "unit day value", which varies according to the scarcity and quality of recreation experiences. (See Walsh, 1986 for a description and appraisal of these approaches). At the time of its introduction, the Unit-day value approach had a role in providing a reasonable number as an alternative to early recreation benefit estimates which were overly large, being based on a total expenditure rather than a consumer surplus concept. Although this might be a suitable shortcut approach in a limited number of cases, at the present state of the knowledge of recreational benefit measurement, it surely is not preferable to an on-site study. Moreover, the Water Resource Council estimates seem to have not been updated for some time, except for adjusting for inflation. Given the large number of empirical studies on the subject which are now published, it seems that a sounder basis is now available. Also, presuming that demands for recreational and aesthetic consumption activities are income elastic, it is very unlikely that relative benefits would stay constant. It would seem desirable to review the foundations of the numbers. The question of accuracy aside, for some of the purposes emphasized in this report-of estimating incremental values of instream water flows or of improvements in water quality-the Unit-Day tables do not purport to address MORE FORMAL APPROACHES TO GENERALIZED VALUES: BENEFIT TRANSFER AND META-ANALYSIS Most studies aimed at estimating nonmarket economic benefits for natural resource and environmental management have been narrowly focused on specific locations or issues. However, with the increased need for economic evaluation, and given the limitations of the Unit Day Value tables noted just above, some have suggested that if generalizable conclusions could be systematically derived from the findings of earlier studies, a less costly approach could be found than performing a new analysis for each case. The process of adapting estimates relating to sites that been studied to sites lacking such studies is sometimes called benefit transfer. This approach is to pool data from already completed studies and subject it to multiple regression analysis. The model could then, in principle, explain the variation in benefits resulting from the differences among explanatory variables 50

69 (subject, of course, to the caveat that the model specification incorporates all relevant variables in the correct functional form). For locations lacking data, the benefits could be predicted by inserting the corresponding values of explanatory variables for those locations into the general model estimated for sites alredy studied. Walsh, Johnson and McKean (1992) represent an example of this approach. The formal quantitative analysis of a sample of research findings with the intent of distilling generalizable conclusions is called meta-analysis. In addition to the study cited above, another example of meta-analysis of nonmarketed environmental benefit studies is found in Smith and Kaoru (1990). One of the most interesting examples of meta-analysis was reported by Carson et al (1996), who tested the ratio of Contingent Valuation (CV) to Revealed Preference (RP) estimates from studies which sought to compare how CV estimates correspond to RP estimates for the same quasi-public good. Their data-base was derived from 616 comparisons of CV to RP compiled from 83 different published and unpublished research efforts for a wide variety of quasi-public goods. Most of the sample studies were environmental goods and services, but also included such issues as job-risk reduction and mosquito abatement. The types of RP studies analyzed included travel cost, hedonic pricing and cost of averting environmental damages. Carson et al reported that the CV/RP ratio to be less than (but close to) 1.0 for each of three different treatments of the data set. These results give support to the proponents of the use of CV methods in nonmarket valuation against critics who contend CV estimates overstate actual willingness to pay. In the remainder of the monograph, the discussion turns to application the methods introduced above to selected alternative water use-sectors. The next chapter takes up the valuation of water in intermediate good production, and Chapters 5 and 6 address consumers goods and public goods. 51

70 4. APPLICATIONS 1: THE CASE OF WATER USED IN INTERMEDIATE GOODS In this and the following Chapters, we turn to discussion of the distinctive attributes of valuing water by individual use sector. The case of water use in intermediate or producers' goods is taken up first. Chosen for discussion here are crop irrigation, industrial water use and hydroelectric power generation. An Appendix to the Chapter briefly introduces some related tools for doing applied water valuation in the intermediate good case. These include short discussions of shadow pricing of capital and labor, and a summary of the process for converting the value of a capital outlay into equivalent uniform annual costs. 4.1 MEASURING THE VALUE OF WATER FOR IRRIGATED CROP PRODUCTION What is referred to as the "value" of irrigation water is a measure of the net economic contribution of water to the value of agricultural production. For a number of reasons, the accurate valuation of irrigation water is of special interest and concern. Irrigation of agricultural crops accounts for the largest amount of water consumed in the world (Gleick, 1993). Therefore, economic feasibility tests of new irrigation projects and of the rehabilitation of existing irrigation facilities should rest on sound economic evaluation. According to the conventional theory of the distribution of factor incomes, the value of a resource is an upper bound on the farmers' ability to pay for water. Empirical estimates of the value of irrigation water provides important evidence on the farmers' ability to pay in implementing cost recovery programs for such development and rehabilitation projects. Another increasingly important use of irrigation water valuations arises in connection with the analysis of economic tradeoffs among water using sectors. Although many agricultural uses of water yield high economic returns, the lowest valued consumptive uses of water are typically concentrated in the production of agricultural crops (Gibbon, 1986), so that intersectoral tradeoff analyses almost always include consideration of the irrigation sector. Yet another need for valuing water for irrigation arises with ground water management. Agriculture is the major source of demand for ground water, and particularly so in regions experiencing overdrafts and other adverse effects of overexploitation. Economic evaluation of groundwater policy frequently requires estimates of the value of irrigation water. This section draws on the conceptual framework developed in Chapter 2, applies it to measuring the value of irrigation water, and discusses the specific problems and pitfalls arising in actual applications. Recall from the earlier discussion in Chapter 2 that the "economic value" of a nonmarketed good or service is defined as the amount a rational, fully informed user would be willing to pay for it. Willingness to pay is formally represented by a demand curve relating the quantities of the good to a series of prices. Conforming to the principle of diminishing returns, the demand curve is downward sloping, reflecting the fact that 52

71 values at the margin diminish as additional units are taken. The demand or marginal value curve is derived from production and applied welfare economics theory (Johannson, 1993, presents a concise and useful discussion of the theory of producers' demand for inputs. See also Just, Hueth and Schmitz, 1982; and Freeman, 1993) Is Irrigation Water Accurately Valued? Confronted explicitly, the problem of assigning an accurate economic value to water in crop production is a complex and difficult task. The overall decision problem is that of a multi-input, multi-output farm household facing production lags, uncertain input and output prices, uncertain production possibilities, heterogeneous labor supplies and alternative employment opportunities for family labor (Huffman, 1992). Irrigated crop production is a dynamic biological process in which input decisions are made sequentially as crops are planted, grown and harvested. Each farmer's decision in this sequential dynamic process is contingent upon results of past decisions, past events, and information regarding future events. Economic evaluation of changes in agricultural production policy implicitly or explicitly requires an accurate predictive model of producer choice. The model should be able to explain observed behavior in specific or similar situations. Then it can be relied on to predict impacts of hypothetical or actual policy initiatives. Models of agricultural production decisions can take several forms, ranging from simple farm budgets (Brown, 1979) to complex mathematical optimization models. Following the usual practice (Hazell and Norton, 1986), it is here assumed that the producer makes choices with the goal of profit maximization, constrained by natural resource endowments (climate, quality and productivity of soils and water), crop production technology, the firm's productive resources (land, labor, capital, water), input and product prices and lastly. political - legal constraints and/or incentives. In the author's experience, estimates of the value of irrigation water are often seriously flawed In terms of direction of bias, the estimates are much more often overstated than understated. The pervasive bias toward overestimates arise from a number of sources, not the least of which is political pressure on analysts to find the "correct" benefit-cost ratio for a project which already has strong political support More mundane sources include excess optimism by the analyst regarding the potential contribution of water-as compared to other inputs such as improved varieties, fertilizer and pesticides, and particularly, human capital-to improved crop production Another source of bias is inadequate consideration of the adverse effect of technological change on the price and value of farm commodities, which has tended to lower the realized return relative to that anticipated in the planning period. Another source of overestimate of irrigation benefits is a tendency towards understatement of the opportunity costs of unpaid familv labor Evidence in support of the claim of bias is the difficulty farmers encounter in repaying costs of irrigation projects (Franklin and Hageman, 1984). While it is clear that self-interest colors many farmers' statements concerning their ability to pay for water, if the benefits were all 53

72 that they are claimed to be, farmers would find it easier to repay at least a small part of the costs of irrigation projects Applied Irrigation Water Valuation: The Residual Imputation Approach and Variations The "residual" method and variation of it are the most common approaches to applied shadow pricing of irrigation water. Recall from the derivation in Chapter 3 that if appropriate prices can be assigned-presumably by market forces-to all resources but one, the remainder of total value of product is imputed to the remaining or residual resource (Heady, 1952). The reader is referred to the previous Chapter, which discussed a number of the general problems and issues arising in residual and related approaches. The pure case of residual imputation (for a single-product farm) is only of interest as a special case. The more typical problem will be to estimate values for a multicrop farming operation. Often, too, the task is to estimate the value contributed by a partial increment or decrement to the water supply, rather than that for the entire quantity. A method developed for such situations is called the "change in net income" approach, to which we now turn The "Change in Net Income" Method Irrigation water is often valued, in real world planning exercises, with a variant of the residual method, the Change in Net Income (CINI) method. The U.S. Water Resources Council (1983) recommends a process which is essentially the CINI method for irrigation benefit evaluation. As shown in the previous chapter, the willingness to pay for an increment of water is the net producer income associated with that increment. Denote Y as a vector of outputs of feasible crops and X as a vector of production inputs, and P represents prices of inputs and outputs. The usual assumption is that farmers are price-takers, and prices are constants. The net income (labeled Z) from producing a given set of crops can be represented by: Z = Yi=. m(yi* x Pyi) - Xj=l...n (Xj x Pxj) (4.1) The Change in Net Income is: AZ = Z 1 - Z4 (4.2) where the subscripts 0 and I refer to the "without project" and "with project" situations, respectively. AZ represents the increment of economic surplus attributable to the project 54

73 or policy-i.e. the measure of benefit that is the point of the effort. The Change in Net Income method is readily adapted to solution on a microcomputer with conventional spreadsheet software. The Change in Net Income approach requires the analyst to make a number of a priori judgments in making the calculations. The most important of these judgments include assumptions for both the with and without situations about a) crop species and acreage of each to be grown, b) the crops' response to alternative amounts and timing of water applied, and c) what irrigation water distribution technologies might be employed. Each of these can significantly affect estimated water use and net income. When the amount of one productive input-such as irrigation water-is changed, the optimal mix of all inputs will also likely change (Johansson, 1993). For example, if faced with reduced amounts of irrigation water, the profit-maximizing farmer would change the amount of fertilizer and other inputs. These adjustments, although not easy to do accurately, can be readily incorporated into the Change in Net Income method. Note also that if land is the only residual claimant in the net income expression (4.1), (as it would be if the without project situation involved rainfed cropping), then (4.2) is the multicrop equivalent to the residual imputation formula given in Chapter 3 above. In other words, Zo represents the opportunity cost of the residual claimant, land in the "without" situation Specific Issues Arising in Application of Residual Imputation to Valuing Irrigation Water Collection of accurate data concerning the situations being modeled is an important concern with this as with any valuation method. Farm surveys are the preferred method of determining the initial "without" situation regarding historical trends in farm size, cropping options, production technologies, input mixes and crop yields. (See Casley and Lury, 1987; or Salant and Dillman, 1994, for discussions of survey research procedures.) However, when time and study resources preclude a survey, secondary data from government, university and extension service reports will have to suffice. Brief "minisurveys" of farmers to verify the accuracy of secondary data may provide a suitable compromise between complete reliance on secondary data and the time and cost limitations imposed by a careful survey. For the "with" cases, forecasts of all the above variables for the relevant planning period must be made. General considerations Although valuing irrigation water is a relatively straight forward task from the accounting perspective, inaccuracies or biases can arise from either or both of the two main elements identified by the conceptual framework: the specification of the physical production function and the pricing of inputs and products. The physical production function identifies the quantities of all necessary inputs and the associated outputs. Pricing of both inputs and outputs involve judgments as to whether market or shadow prices should be 55

74 used for each input and output, and if shadow prices are necessary, how they are to be derived. (Brief introductions to the shadow pricing of important inputs, including capital and labor, are given in the Appendix to this Chapter). Biases can be introduced by any of several types of errors or omissions. The residual technique is extremely sensitive to small variations in assumptions about either the nature of the production function or about prices, so the approach is most suitable for cases where the residual claimant (in this case, water) contributes a significant fraction of the value of output. Omitting an input that should be represented in the production function will permit the contribution of that input to be attributed to the residual claimant, in this case, irrigation water, thereby overstating the value of water. A similar but lesser bias arises if the amount of input, such as fertilizer, is merely underestimated rather than omitted entirely. Other biases arise from improper pricing of inputs or outputs. An overpessimistic price for a commodity results in a residual that is too low, although the opposite error of overoptimism is more frequently encountered. Over- or underpricing inputs induce corresponding but opposite effects on the estimated value of the residual. For example, a zero shadow price for family labor, practiced by the U.S Bureau of Reclamation and once recommended by some academic analysts, has the effect of attributing the productivity of labor to the water resource. In actual applications of this model to crop irrigation, a number of specific issues must be addressed. Estimating the value of irrigation water, in practice, usually involves the long run forecasts of the local agricultural situation "with" versus "without" the increment of water under study. These forecasts must be translated into specific values for input prices and quantities and product prices and yield increments. Specific issues which must be dealt with in application include questions such as whether nonwater inputs (such as wages or interests rates) must themselves be shadow-priced, and if so, how? Other practical questions include which crops to include in the "with" project situation, and what is the increment in yield attributable to the added water supply? Other problems are how to forecast technological and price changes over the life of a project, and changes in cropping For example, in crop irrigation project planning, which often involves long (fifty years or more) time horizons, analysts often include a forecast of technological improvement in productivity over the life of the project Errors may be introduced either by overestimating the actual technological progress which actually occurs, or by underestimating the corresponding nonwater input levels (e.g for pesticides, fertilizers, human capital in the form of expertise) which are necessary to achieve the predicted level of output. A related point is seen in the historical tendency throughout the world for real crop prices to decline with technological improvement. Therefore, if improved yields are forecasted, the corresponding adverse effects on product prices should also be considered. [The author has reviewed project plans in which yields were forecasted to increase at a constant percentage annually, but input costs and commodity prices were assumed to remain constant over the life of an investment. Such a practice is a way to boost the chances of economically justifying an investment, but is unlikely to be a correct approach to evaluation (Young, 1978).] 56

75 Short-run or long-run models? The model of farmer decision context is important in irrigation benefit estimation. The omission-of-variables problem often occurs when a short run modeling framework is used to represent a long run planning situation. In short run contexts, it is appropriate to treat certain production inputs as fixed, and to omit charges for those fixed inputs in calculating the residual value. Water project analysts who were trained in traditional agricultural economics methods may incorrectly apply these approaches to irrigation benefit estimation. For example, a typical farm management planning exercise is to determine the annual mix of crops which will maximize returns to fixed family resourcesland, water, labor, management, etc. 'This is a short run planning problem, but if applied to measuring irrigation benefits, would lead to imputation of the values of all fixed resources to water. Although short run models have their place, the long run is appropriate in most water development and allocation planning contexts, and in those cases all factor inputs should be treated as variable and costs assigned to them. Evaluating proposals to increase water supply for crop irrigation encounters examples of several of the production function and pricing issues discussed above. The prediction of input use, production technologies and yields over a long planning horizon is a key aspect of the irrigation benefit estimation process. These elements must be treated in a coordinated and consistent fashion. The water-crop production function A production function, which mathematically or graphically represents the relation between inputs and output(s) in a production process, serves as a basis for describing, explaining and predicting the output expected from a specified level of inputs. For irrigation water valuation, water-crop production functions play an important role, serving as building blocks for models of farmer response to alternative water management policies. Production functions for irrigated agricultural crops can be derived directly from experiments, from statistical analysis of secondary data or indirectly by mathematical simulation models (whose parameters are normally obtained from direct observations). Reviews of early approaches to developing water-crop production functions are found in Carruthers and Clark (1981) and Vaux and Pruitt (1983), while Boggess, Lacewell and Zilberman (1993) address the more recent literature. Many analysts prefer to rely on experimental production functions because they are deemed more realistic and reliable. Hexem and Heady (1976) described studies of experimentally based water-crop production functions. However, experiments are expensive and time consuming, and may not be readily generalizable beyond the local experimental conditions. Water-crop production function analyses based on secondary data are few in number, because of the scarcity of appropriate data. (See Moore, Gollehon and Negri, 1993, for an example.) While valuable for broad scale testing of hypotheses of interest to policy makers, in contrast to experimental or simulated water-crop production functions, those based on secondary data are usually aggregated over crops and production regions, and are seldom suitable for disaggregation into other planning models. 57

76 As more precise scientific knowledge of the factors affecting crop growth become available and the costs of computing fall, simulated production functions become a more feasible approach in many situations. Crop yield as a function of applied water is often simulated with models which compute yields from total water applied during the season. Simulated production functions can be readily adapted for specific soil and climatic conditions, and so provide a flexible and relatively inexpensive method of producing agricultural production functions for varying local conditions. Economists and agronomists have worked separately and jointly on developing simulation models of irrigation crop response. (For discussions of the particular issues arising in simulating water-crop production functions, see Just, 1991; Cardon and Letey, 1992a, b; and also Letey and Dinar, 1986, and Dinar and Letey, 1994.) Measuring nonwater inputs At first look, an accurate estimate of the inputs required in the with and without cases is straightforward. However, in residual analysis applied to irrigation, some tendency has been observed toward omitting or undercounting inputs, rather than erring in the opposite direction. (See Ahearn and Vasavada, 1992; and Eidman, et al., 1994, for recent comprehensive reviews of the conceptual and measurement issues encountered in calculating costs and returns in agricultural production in the U.S.) It is convenient to separately discuss issues arising with nondurable capital, durable capital and labor. Considering first the nondurable capital inputs such as seed, fertilizer and pesticides, here little problem is encountered. Agricultural universities, local extension offices and government research agencies can usually be drawn upon to provide satisfactory approximations to prospective nondurable input use levels in the with and without irrigation cases. Turning to labor and the human effort, the same sources as mentioned above for other inputs can provide estimates of labor input coefficients. A major issue, in the author's view, is the underreporting (and incorrect pricing ) of "management" and other non-manual labor inputs. If labor is accounted for only in terms of the hours expended on performing specific field work tasks associated with land preparation, planting, weeding, irrigating and harvesting, an important part of the production picture has been missed. Even in developing countries, specialized work activities are required to make production choices, input purchases and product marketing decisions and to learn about potential new technologies. In the related context of accounting for agricultural productivity increases at the national level, T.W_ Schultz (1981; cited in Hallberg, 1992) has identified a tendency toward underreporting of specialized labor and human capital inputs and the consequent overattribution of the contribution of inputs such as irrigation water. Management effort is not easy to quantify (Burrell and Hall, 1980). The most frequently employed method is an opportunity cost approach, which rests on the rates charged by commercial farm management services to perform managerial tasks for absentee owners. The fees charged typically reflect the complexity and riskiness of the crops to be grown, and are a fixed percentage of either variable production costs or total revenues. According to Willet, (1992), rates of five to ten percent of either operating costs or total 58

77 revenues are representative figures for the United States. The U.S. Bureau of Reclamation suggested "at least six percent of variable production costs". The writer is of the opinion that for field crops, 5% of gross revenues is an appropriate figure for management and nonfield labor and that higher percentages (e.g. 8-10% of revenues) are warranted as management charges for perishable specialty fruit and vegetable crops. Measuring outputs Three issues of measurement arise on the output side of the equation. One is establishing which crops are likely to be produced under both the with and without irrigation conditions. Second is the basic level of yields relative to assumed input levels, and the last is correctly identifying the incremental contribution of irrigation water to production. Specialty crops On the lands of any irrigation project, numerous crop alternatives are possible. The planning process must predict with reasonable accuracy what crops are to be grown in the with and without irrigation situations. Further, the proportion of acreage devoted to each crop must be specified. Proposals for irrigation projects often incorporate a substantial component of specialty crop production. Specialty crops (which here will refer to perishable vegetables, vine and fruit crops) appear to generate a high return over operating costs and therefore yield high net returns to water. However, two opportunities for overestimation of returns to water arise with specialty crops. First, the return over operating capital costs per unit land (hectare) must be carefully allocated. In particular, the high profits must be examined carefully. They are not so much a signal of scarcity of specialty products and the resources to produce them, but a sign of scarcity of the technical, managerial and entrepreneurial skills required in successful specialty crop production. High returns represent a return more to the high human capital requirements and to compensate for the market and production risks inherent in perishable crop production than to the water resource. In order to avoid overestimation, an imputation to risk-bearing and management are needed (Burrell and Hill, 1980). The second opportunity for overestimation of returns to water is the proportion of acreage devoted to specialty crops in the long term development plan. Such plans may assume a much larger proportion of resources devoted to specialty crop production than historically has been observed. In actuality, in most nations, only a small proportion of lands are actually devoted to specialty crop production. [For example, in the state of California in the U.S., which is renowned as the nation's vegetable basket, vegetable and noncitrus fruit production accounted for less than twenty percent of total agricultural land use in the state (U. S. Department of Agriculture, Agricultural Statistics, 1994) Accordingly, in few nations is there reason to assume shortages of fruits and vegetables that government investments are needed to alleviate. In most cases, any specialty crop production which occurs on a new irrigation project is likely to merely have displaced production elsewhere in the nation. 59

78 A review of the manuals of procedure for water planning does not yield a commonly agreed upon approach to the question of how to treat specialty crops. Neither those originating primarily from Europe (Bergmann and Boussard, 1976; OECD, 1985) or from the World Bank (Gittinger, 1982) appear to regard it as a problem (although neither of the latter two volumes was presented at the level of detail adopted here). In the U.S., the Water Resource Council in the 1970's took the position that specialty crop acreage on new irrigation projects would merely displace similar crops on already producing lands (irrigated or otherwise) elsewhere in the nation, and that there was no reason for government investments to augment specialty crop production. Therefore, the Council's Principles and Guidelines (1983), continuing the practice of earlier manuals, prohibited any incremental benefits being attributed to specialty crops in new irrigation projects. Although the present writer does not suggest that specialty crops be always eliminated from calculation of irrigation benefits, the matter clearly does require careful attention on a case by case basis. First, it must be demonstrated that no displacement of other producers in the nation is likely. ( It may be that the new development has a location advantage-in terms of transportation costs or productivity-in which cases, the reduced costs of production and marketing would be a real welfare gain.) Second, as indicated above in the discussion of labor resource valuation, to avoid erroneously attributing labor efforts to the residual water resource, the writer recommends that a contribution of owner/operator managerial efforts to crop production be recognized and priced at rates higher than the going wage field labor. In view of the difficulty of establishing prices and quantities for management, the opportunity cost approach of assigning a percentage (8-10%) of gross sales to management for specialty crops appears to be an appropriate compromise. Further, the amount of land area (in hectares) assumed to be devoted to specialty crop production in plans for new irrigation projects should not exceed the proportion existing in the without project situation. Should livestock production be counted in analysis of irrigation benefits? A question related to the inclusion of specialty crops is whether to include returns from livestock and poultry operations in the analysis. Farms with irrigated cropping activities also frequently engage in livestock production, relying in part on forages and feeds produced on the farming operation. Addition or augmentation of irrigation water supplies produces more feeds and forages, which may find a better outlet through farm livestock and poultry feeding operations than on the open market. Correspondingly, irrigation project planners often choose to incorporate livestock activities into their analysis. The recommendation here is that livestock not be included in the budgeting. A project providing an increment to irrigation water supply directly augments only feed and forage production, not livestock production. By importing feed into the project region, livestock could be grown in the absence of the project. The livestock enterprise can be best regarded as a form of secondary processing for the feeds and forages, rather than a primary enterprise directly impacted by the irrigation water supply. Introducing another intermediate good enterprise into the analysis opens the way for additional errors in imputing benefits to incremental water supplies. 60

79 It is probably simpler and more accurate to directly price the incremental forage than to impute a price through the livestock enterprises. Livestock production is, as is cultivation of specialty crops, often more demanding of managerial and entrepreneurial skills than is regular crop production. As discussed above, it is difficult to accurately measure and price these inputs. Moreover, under competitive market conditions, the price differential between livestock sales (meat, milk or fiber) should reflect the costs of inputs plus the opportunity costs of management and risk-taking needed to attract resources into livestock production. However, markets usually exist for feeds and forages which can avoid the potential problems. Feed grains used in livestock production can be valued at cost of acquisition. If they are scarce in the project region, the market price will necessarily reflect the cost of importing from grain surplus localities. Fresh forages used for livestock grazing may appear to be difficult to price; however, rental markets for grazing by the week or month often can be identified. In such cases, it is possible to use the rental rate to price the increased forage outputs. Specifying the production function with versus without the project Issues relating to the production function for irrigated crops can be conveniently further divided among a) non-durable capital inputs, b) durable capital inputs, c) land, d) labor, and e) general overhead and management (including taxes, insurance). Non-durable capital inputs In agricultural production, the main nondurable inputs are seed, fertilizer, pesticides and energy (fuel and electricity). (Labor is discussed separately, below). The main concern for water valuation is to assure that all inputs actually expected to be used are included and in the correct quantity. The use of a Table of Operations and Inputs as described in the following section of this report provides an explicit accounting of input factors, which when reviewed and approved by experienced technical specialists, provides suitable assurance of completeness and accuracy. As noted above, omissions or underrepresentation of nonwater inputs lead to overestimation of the residual value of water, while listing of unproductive inputs has the opposite effect. Perennial tree and vine crops (as well as certain long-lived forage and pasture crops) present an issue of accounting for initial investment in establishing the orchard or crop. If full detailed year by year budgets are not performed, some way of amortizing the fixed cost of establishment must be provided in the annual cost and return budgets. Durable capital inputs Durable capital inputs present somewhat more difficulties than do nondurables. Items such as tractors and their associated tillage and other equipment, harvesters, and buildings present special difficulties. The unit budgets usually represent one year's operations; annual depreciation and opportunity interest on the one year's portion of the life of a durable item of equipment must be quantified and costed out. Burt (1992) reviews the 61

80 major theories of depreciation and opportunity interest rates. Depreciation and opportunity interest rates should be in real, not nominal terms. A frequent error associated with costing durable equipment committed by nonspecialists is using the equipment item's expected life, rather than the project planning period as the time period over which the item is depreciated. One of the rules of cost-benefit and other forms of discounted cash flow analysis calls for a comparable time period for all durable investment items (James and Lee, 1970). In the frequently encountered case where the economic life of a durable item, say a sprinkler system or an improved irrigation ditch is less than the project life or planning period, the present value of the cost of replacement of the durable items throughout the planning period should be added onto the cost of the initial item before depreciation is calculated Implementation of the "With versus Without" Test in the Residual and Change in Net Income Methods Implementation of the analysis requires the assembly of farm budgets for the with and without cases. Farm budget analysis draws on agricultural production expertise, economics and accounting to infer returns under alternative situations. Analysts prepare budgets for the purpose of identifying and valuing the inflows of resources and the corresponding outflows of products within a specified accounting period, usually one year. A budget provides the basis for assessing the impact of a specific plan, in which a clearly specified package of inputs is committed and from which emerge a flow of costs and revenues over the project life (Brown, 1979; Eidman, et al. 1994). Farm budgets can be used for many purposes, including land purchases, choice of production technology as well as the topic of incerest here, planning for crop irrigation decisions. One form is the partial budget, which analyzes changes which have a limited, short run effect on the total farm organization of resources. A "complete" budget is appropriate for proposals which have major, long term impacts on resource organization and income. Water planning activities nearly always require the complete budget approach. A "representative farm model" is a helpful analytic device for portraying the farm situation being analyzed. Because it is impractical to evaluate each individual farm holding, the simplification to one or more representative farm models is common practice. The farm model begins with a list of the assumed characteristics of the farm or farms being evaluated. An inventory of the principal resources defining a representative farm comprise the representative farm model. These can include the land, including productivity of soils, labor supply, climatic factors, financial status, machinery and equipment inventory and perhaps buildings. A frequent practice is to identify several representative farms for analysis, based on main production emphasis (field crops, perishable crops, livestockbased), resource productivity (such as that due to variations in soil quality or microclimate), managerial capability or financial strength. Farm models also involve listing of the production options available to the producer, including the range of feasible crop and livestock enterprises and technological 62

81 options for production. The model should be based on realistic assumptions about productivity of resources, markets available to dispose of products and managerial capability. The Unit Table of Operations and Inputs. The starting point for detailed and explicit farm budget analysis is the unit table of operations and inputs. They are designed to represent the technical and economic opportunities for producing a technical unit of a farm enterprise, and provide a common format for display of the necessary assumptions regarding production of agricultural enterprises. Technical units are usually land units (hectares or fractions thereof, etc.) or animal units (one head of livestock). For the purposes here of analyzing irrigation decisions, a land measure is the appropriate unit. A "Table of Operations and Inputs" for each crop makes explicit each step required for crop production, its resource requirements and the resulting outputs. Table 4.1, patterned after those in earlier textbooks-e.g. Hedges (1963) provides an example. The assumptions necessarily correspond to an assumed inventory of machinery and equipment (appropriate to the size and financial status of the representative farm) which a farmer can call upon to produce the crop in question. The next step is to assemble a Unit Crop Budget to calculate and display the net returns over variable costs per unit land for each crop option. Table 4.1 illustrates graphically the considerations which enter into the Unit Crop Budget. The Total Farm Budget. Illustrated in Table 4.2 is the Total Farm Budget, which combines the information in Unit Crop Budget to determine, first, Farm Income Net of Variable Costs, and second, by deducting fixed or overhead costs, deriving Net Return to the residual claimant(s). In Table 4.2 are shown the Total Farm Budgets for the "with" and "without" situations. After the "Change in Net Income" is calculated (Line F), the final "bottom line"-estimated net benefit per unit water-is imputed in lines H and J. 63

82 Table 4.1 Representative Format for Unit Table of Operations and Inputs (One Hectare for One Crop) Crop.. Projected Yield per Hectare Machinery Per Hectare In uts Power Source Equipment (size & (size & Machine Labor Materials Operation type) type) Hours Hours (Itemized) Seedbed Preparation l Step I Step 2 Step 3 Step 4 Plant (seed) Fertilize (fertilizer) Pesticide (chemicals) Cultivate Irrigate (water, cubic meters Harvest (bags, ties) Haul (miles) Store (months) 64

83 Table 4.2 Representative Total Farm Budget Format for With and Without Analysis of Irrigation Developments Part I - Return Over Variable Cost by Crop "Without" Development "With" Development Situation Situation Item Crop A Crop B Crop C CropD $ $1 1$ A. Revenues Per Hectare 1. Projected Yield/Hectare 2. Projected Price/Unit 3. Projected Revenue/Hectare B. Variable Costs Per Hectare 1. Land Preparation 2. Plant 3. Fertilizer and Pesticide 4. Other Pre-Harvest Operations 5. Irrigation 6. Harvest i X 7. Hauling and Storage 8. Management Charge 9. Operating Interest 10. Total Variable Costs r: X X X X C. Total Return Over Variable Cost: "Without and "With" Development 1. Return Over Variable Cost/Hectare 2. Hectares 3. Total Crop Return Over Variable Cost 4. Total Farm Return Over Variable Cost 65

84 Table 4.2 (Continued) Representative Total Farm Budget Format for With and Without Analysis of Irrigation Developments Part II - Net Return to Farm Operations "Without" "With" Development Situa- Development Situa- Item tion tion $ $ D. Annual Overhead and Annualized Capital Costs (Total Farm) 1. Land Development 2. Machinery and Equipment 3 Buildings 4. Transport 5 Irrigation Water Supply 6. Irrigation Water Distribution 7. General Overhead (Taxes, Insurance, Office) 8. Total Farm Capital and Overhead Costs (Sum of 1-7) E. Net Farm Income (C3 - D8) F. Change in Net Income ("With minus "Without") G. Cubic Meters of Water Withdrawn H. Net Benefit per Unit Water Withdrawn ($/Cubic Meters) I. Cubic Meters of Water Depleted J. Net Benefit per Unit of Water Depleted ($/Cubic Meters) Part III - Change in Net Income Calculation 66

85 To insure realism, operations and input tables such as that displayed in Table 4.1 are best developed in a collaborative process with plant and soil scientists, agricultural engineers and especially with local agricultural extension specialists. (Note that these tables reflect developed country technology and pricing. For developing country conditions, suitable eliminations or adjustments can be made as appropriate to the situation) Mathematical Programming Models for Irrigation Water Valuation What is essentially is the Change in Net Income method can be adapted to mathematical programming models (Williams, 1990) of farm situations (Hazell and Norton, 1986; Rae, 1994) to yield an estimate of the functional relationship between net benefits and irrigation water use. Mathematical programming has been adapted to irrigation water valuation over the past several decades as the technique itself and as computer technologies have advanced. For most water planning activities, the mathematical programming approach to modeling irrigator decisions is much more demanding of analytic training and resources than most planning teams can be expected to possess. However, with the rapid extension of desk computer capacity and the skills and training of evaluation staffs, the technique will increasingly be adapted to real world planning activities. Hence a brief review of the approach is offered here. (See Chapter 3 for an overview of the basic characteristics of mathematical programming models) The basic Change in Net Income approach requires the analyst to make a number of a priori judgments about a) crop species and acreage of each to be grown, b) the crops' response to alternative amounts and timing of water applied, and c) what irrigation water distribution technologies might be employed. A more realistic model of farmer behavior would make these considerations internal to the model. Analysts who wished to introduce farmers' choices regarding which crop to produce, water application rates and production technologies as decision variables in their models were quick to take advantage of mathematical optimization techniques, such as linear or quadratic programming. An early variant of the programming approach was Anderson (1968), who while retaining the fixed crop acreage assumption of the whole-farm budget approach, utilized computer simulation to facilitate representation of multi-stage crop response to alternative amounts and timing of water application in a model of an irrigation delivery system. Numerous applications of linear programming to irrigation planning followed, only a few of which can be mentioned here. Early models provided only for omission of marginal crops in response to increased price or scarcity. Extensions of the mathematical programming approach to irrigation proceeded along several different paths. Sequential or multi-stage decision processes and crop response to varying water application rates were one avenue. Young and Bredehoeft (1972) took this direction, and also showed that the water application portion of the Anderson (1968) simulation model could be more easily and accurately represented by a linear program. Other extensions included representation of seasonal crop response to water (based on highly detailed agronomic simulations) and irrigation application technology (Bernardo, et al., 1987). (See the discussion of water-crop production functions earlier in this section). 67

86 The usual approach is to formulate a programming model of a representative farm situation specified to maximize net return to the residual claimant (the water resource in this case) subject to constraints on water and other resources. The model is solved for each of a number of increments of water supply and the net return to each increment of water derived from the incremental change in the objective function (Burt, 1964; Bowen and Young, 1985; Bernardo, et al., 1987). The objective function value for each solution of the model provides an estimate of the value of water for the supply scenario assumed for that solution. The marginal benefit function can then be determined. An alternative approach is to trace out a derived demand function by solving the model for a range of water prices and recording the corresponding optimal water use rates. Formulation and solution of a farm model on a microcomputer provides the opportunity to increase the degree of detail in representing the farmer decision maker, incorporating a wide range of technological production options, input levels and crop choices as endogenous to the model solution. The tradeoff is that the programming approach is more demanding of analytic resources. More specialized computer hardware and software and a higher level of staff training are needed to solve programming models, and the additional detail calls for empirical specification of correspondingly more cropping and irrigation activities. To insure accuracy of the results, the same attention must be paid to the empirical assumptions and procedures recommended for the basic residual techniques. Most examples of mathematical programming to the study of irrigation demand and benefits have been partial equilibrium, deterministic and static. Extensions from this base include Howitt, Watson and Adams (1980), who modeled irrigation decisions with quadratic programming to allow crop prices to vary with regional output of irrigated crops. Berck, Robinson and Goldman (1991) present an even more theoretically sophisticated regional economic model for water policy analysis, a computable general equilibrium model of agricultural water use in an important California agricultural area, the southern San Joaquin Valley. Based on the Leontief input-output format, this CGE model incorporates irrigation water as an important factor of production. In contrast to the fixed price assumptions of input-output models, product demand functions are incorporated into the CGE model. By accounting for value-added elements as opportunity costs, the approach taken by Berck, et al. appears to overcome the potential shortcomings of the value-added method criticized above in Chapter 3. Optimization models of intersectoral regional water allocation by Vaux and Howitt (1982) and Booker and Young (1994) incorporate demand functions for water solve for optimal prices of water in a regional or basin-wide context. Taylor and Young (1995) developed a discrete stochastic sequential programming (DSSP) model of sequential uncertain multi-crop production process characteristic of irrigated agriculture. They show that benefits increase with increasing reliability of water supplies. Another path was to utilize dynamic programming to represent the sequential choice problem faced by irrigators in the presence of limited water supplies. Representative of this approach is Dudley's (1988) sophisticated analysis of optimal land and water use on irrigated cotton in Australia. Knapp and Wichelns review the dynamic optimization approach with extensions to water quality and drainage. The dynamic programming approach provides a rigorous representation of the problem of sequential water-use decisions in the face of uncertain water supplies. However, it sacrifices some realism because its 68

87 heavy computational demands have limited the analysis to considering but one crop at a time. After this lengthy exposition on the major residual-based methods of valuing irrigation water, we briefly discuss two other techniques which are sometimes employed, the hedonic and alternative cost approaches The Hedonic (Land Value) Approach Applied to Crop Irrigation What is sometimes called the "land value" approach in irrigation water valuation is an example of the hedonic method discussed in Chapter 3. If agricultural land and real estate markets in the study area are active and competitive, then a comparison of farm land values derived from samples of land sales representing irrigated (or partially irrigated) and nonirrigated production can provide a useful and relatively convincing revealed preference approach to valuing irrigation water. Many economists are more comfortable with resource values that are derived from actual market behavior of buyers and sellers than in the calculations of analysts with no financial stake in the accuracy of the calculations or in the outcome of the investment. Analyses of sales values is an example of the hedonic property value model. See Freeman (1993, Chapters 4 and I 1) for a detailed exposition of this approach, including methods of deriving information on preferences from property values and on deriving measures of benefits. In what follows, we present a simplified approach to land value techniques as applied to valuing the economic contribution of irrigation. The earliest examples of hedonic methods applied to irrigation water valuation known to the author were by Milliman (1959) and Hartman and Anderson (1963). Their work anticipated most of the major developments of the hedonic method in the 1970's but predated the coining of the term "hedonic". From a list of several years of actual sales of irrigated farms in northeastern Colorado, Hartman and Anderson collected data on farm sales prices, the estimated amount of water rights, irrigated and nonirrigated acreage of cropland, and other factors hypothesized to influence farm land sales value, such as capital improvements. (The fact that holdings of water rights relative to land area varied significantly across the regional sample was key to the feasibility of the method in this case). Regression analysis of the attributes on sales prices yielded a coefficient for each attribute. The coefficient on water rights holdings was interpreted to derive the incremental value of a water right. A more recent example applied to irrigation water value is Torell, Libbin and Miller (1990), who compared sales of irrigated and nonirrigated lands, measured the value of ground water for irrigation in the southem High Plains in the United States (in Colorado, Kansas and New Mexico). The land value approach to irrigation water valuation is usually based on comparisons of the selling prices of irrigated lands with the prices of nonirrigated lands. The buyers and sellers in the land market are assumed to be rational and fully informed. Assume further that the observed land prices represent the amount a fully informed buyer would pay for the rights to a stream of net rents from land ownership, which is the discounted present value of that stream of net rents. The data from both alternatives (with 69

88 and without irrigation) must be reasonably comparable in all respects (soil quality, climate, etc.) except for the availability and application of irrigation water, or there must be data from which the effects of these factors can be controlled for in the statistical analysis. Estimated land values can also be obtained in other ways, including self-reporting of the value by owners, and estimates by professional appraisers, but actual market transactions are to be preferred. (The U.S. Water Resources Council's (1983) Principles and Guidelines manual seems to suggest that professional appraisers are likely to be the most appropriate source, but the availability of such specialists may be limited in many instances.) Imputing a value of water via the land value approach requires at least three steps. First, from observations on land market transactions, the difference between prices per unit area (e.g. per hectare) of land with irrigation and those without irrigation is determined. The difference can be derived from means of observations on land transactions. However, because transactions prices usually differ according to a number of other factors, ideally the net contribution of water is estimated via statistical regression analysis, which controls for variations in any factors other than the existence or nonexistence of irrigation. (For example, Crouter, 1987, Xu, et al., 1993; and Torell, et al., 1990 all included among their independent variables a list of other variables, such as investments for improvements and irrigation system, soil quality, distance to urban centers or town, and size of parcel sold.) Next, the amount of water typically available each year to the irrigated land must be estimated. This amount of water is divided into the price difference obtained in Step 1 to derive an estimate of the gross unit value of water. Because the gross unit value from Step 2 represents a capitalized stream of net rents, in the typical water planning exercises it must be converted (Step 3) to an annual equivalent value. Step 3 is accomplished by inserting the unit value difference, an appropriate interest or discount rate and a planning period assumption into the present value expression and solving for the annual equivalent unit value of water. Several specific concerns must be addressed to arrive at a suitable estimate of the unit value of water. One problem is that several years of market observations may be necessary to arrive at an average land value. In that event, the observations should be adjusted to account for any general inflationary trend in land prices, so that all price observations are on a common real price basis that is comparable to cost estimates to which the benefits are compared. The incremental value derived at this stage is, of course, that of a capital asseti.e. it represents the discounted present value of a stream of forecasted annual incremental values. Hence, to convert the difference in land prices to an annual value of the type usually used in cost-benefit analysis (and comparable to that derived by the residual method) both an interest or discount rate and a planning period must be selected, and an assumption made how the (unknown) annual values behave over the years of the planning period. For land asset purchases, a long planning period is usually assumed. Most analysts settle for the simplifying conventions of the farm appraisal profession, which assume land is a perpetual asset (purchases have an infinite life) and that the annual income are constant over the planning period. For a constant interest rate r, as the length of the planning period approaches infinity, it can be shown that the present value of a time 70

89 stream of constant annual values (denoted A) approaches A/r. The capitalization formula in this special case reduces to: PV = A/r (4.3) Recall the assumptions of the model in (3). namely that the potential investors and potential sellers (present owners) estimate the future streams of net incomes provided annually by agricultural production (with and without irrigation water) and discount it to determine the willingness to pay for and willingness to sell agricultural land. The market for agricultural land is assumed to be in equilibrium Consider now several analytic issues regarding implementation of this technique. First, if the analysis requires the value be expressed in annualized (rather than in present value) terms, what interest rate does the market use to discount future returns? Should the analyst employ the market rate or a social rate of discount to convert capitalized values into annual values? Should it be a real or a nominal rate? Should a premium be added to reflect riskiness of the investment? Second, regarding the estimates of future income streams: the market may be assuming a growth in land value due to general price inflation or to localized demand for nonagricultural land uses. In nations with a history of general price inflation, irrigated land is likely to be seen as safe store of value, and price of irrigated land may be larger than could be reasonably inferred from the capitalization of current net incomes. If any of the market value attributable to potential real appreciation from nonagricultural (residential, industrial or recreational) demands for the land, this factor should somehow be accounted for by deducting the premium attributable to potential nonagricultural demand for land from the imputed irrigation gains. The hedonic approach is likely to have a relatively limited potential for estimating the value of irrigation water. At least, it should not be the only basis for benefit estimation. This conclusion will be particularly true in developing country contexts. In addition to the assumption of a properly functioning real property market with frequent transactions, the data requirements and statistical techniques will challenge the time and budget limits faced in real-world project planning. Even academic researchers with adequate research resources have often found it difficult to derive statistically significant estimates of the contribution of water supply to land market transaction prices, even after controlling for other factors such as investments in irrigation system, farm buildings and improvements and distance to the urban centers. Nevertheless, as a gross measure of benefit, land market prices can be quite helpful in forming a judgment of the contribution of water and the likelihood of a project's economic feasibility. At the very least, if the observed differential between market prices per hectare for irrigated lands and nonirrigated lands are considerably smaller than the public investment per hectare required to provide irrigation water, this provides a warning flag that estimates of project benefits must be examined skeptically. 71

90 The "Alternative Cost" Approach to Irrigation Water Value As discussed above in Chapter 3, another method appropriate to evaluating intermediate goods is the alternative cost approach. When estimates of a direct demand schedule proves difficult because of lack of data or other reasons, the alternative cost approach may be a solution. The alternative can be any feasible alternative for accomplishing the project purpose. The main limitation of using the technique is that the alternative itself must be economically feasible; if not, then any scheme, whatever the cost, could be used to claim justification. Alternatives can be private sector activities, or projects potentially undertaken by other public agencies. In any case, estimates of irrigation water value should be tested against the cost of the most likely alternative method of supplying water to assure that it is the least cost alternative. Estimated willingness to pay for project water should not be asserted to exceed the cost of the least costly feasible alternative. If the cost of the alternative is less than the estimated direct benefits of the project in question, the cost of the alternative should be used as the measure of benefit. For example, where hydrologically, technically and institutionally feasible, ground water costs might be used as an alternative to surface water supply. However, only if ground water is in plentiful supply, will this be a straightforward exercise in calculating costs of ground water extraction. To conventional ground water extraction costs might need to be added external (uncompensated third party) costs. These might include effects from over-rapid extraction, leading to water table decline, with attendant pumping cost increases. Ground surface subsidence, leading to structural damages on the surface, or intrusion of poor quality water are other possible types of external costs (Young, 1992). Assembling complete and accurate estimates of external costs from pumping are likely to be relatively difficult for project planners Measuring Benefits of Improved Quality of Irrigation Water The major water pollutants affecting irrigated crops are dissolved mineral salts ("salinity") including mainly chlorides, sulfates, nitrates and bicarbonates of sodium, calcium, potassium and magnesium. Mineral salts, dissolved by water from soils and rocks, are found to greater or lesser degrees in all surface and ground waters. When water is applied for irrigation, these minerals become concentrated in irrigated soils when part of the applied waters evaporate through plants, and adversely affect crop productivity. Although the effects of the particular ions on crop productivity vary, the usual approach is to lump all salinity into a macro measure "total dissolved solids (TDS). Crops vary in their sensitivity to salinity. Generally speaking, the least sensitive crops are also the least valuable, so areas irrigated with highly saline waters tend to emphasize low-valued types. Due to the concentrating effects of irrigation, the most serious salinity problems are often found downstream from irrigated areas. Farmers, to avoid crop damage from 72

91 the salts, apply water in excess of crop needs to dilute the salts and to drive them below the plants' root zone. These concentrated mineralized waters drain into either subsurface or surface watercourses and may be transmitted to subsequent users of the waters, adversely affecting productivity and usefulness. Economic benefits of salinity abatement can be estimated by the change in net income procedure. The approach is similar to that described earlier for changes in quantity of irrigation water supply. In this case, the increase in crop yields from reduced salinity will bring about an increase in net income that measures the benefit of salinity reduction. Production functions which reflect the change in yields due to reduced salinity have been studied extensively. [See Letey and Dinar (1986); Letey, Knapp and Solomon, 1990; and Cardon and Letey, (1992)]. Net producer income is also affected by the response farmers make to increasingly saline waters. Farmers can adopt several approaches to coping with increasing salinity in irrigation water. The major adjustment is to shift to more salt-resistant, but usually less profitable crops. If water supplies are adequate, more frequent irrigations can leach out salts from the root zone. High efficiency irrigation methods (sprinklers, drip irrigation) can offset water shortages to the same effect. Each of these technologies can be built into either budgeted or mathematical programming representations designed to estimate the change in net income due to salinity reduction. The mathematical programming technique can readily reflect the range of irrigator responses to salinity-changing to crops of different tolerance; changing the number, timing or amounts of irrigation water applied; and changing water application technologies. Beginning with Moore, Sun and Snyder (1974), most subsequent efforts at measuring benefits of salinity reduction have adopted the mathematical programming approach, (e.g. Booker and Young, 1994). 4.2 VALUING WATER IN INDUSTRIAL USES Water value assessment in industrial uses is complex, first, because there are about as many types of water uses as there are industries, second because it is often difficult to locate the exact data necessary to estimate demand equations and third, because deriving values or shadow prices requires steps beyond industrial water demand equations. A major survey of issues in industrial water demand analysis is presented in Kindler and Russell (1984a), (e.g. Kindler and Russell, 1984b, Stone and Whittington, 1984; Russell, 1984). However, the main concern in that volume is with the effects of price and other factors on water use. The related issue that concerns us here-of deriving shadow prices for water in industrial use-is given relatively little attention. In the developed nations, most industrial water use is for cooling, mainly of thermal electric power plants (Solley, et al. 1993; Gleick, 1993; Rogers, 1993). Manufacturing and processing industries, such as chemical production, petroleum refining and pulp and paper production an food processing, constitute another class of uses. Cleanup, sanitary and fire protection account for most of the balance of industrial water uses. 73

92 Two broad approaches to modeling industrial water demand and value are available to economic analysts (Kindler and Russell, 1984b, Stone and Whittington, 1984). As discussed in Chapter 3, the main feasible alternative approaches to valuing water as an intermediate good include an econometric or statistical demand analysis or a mathematical optimization modeling method (similar to the residual method discussed for irrigation) The Econometric Approach to Industrial Water Demand Analysis The statistical or econometric approach involves making inferences from actual observations on quantities consumed and costs of water, together with the corresponding data on other explanatory variables. An abstract demand function is formulated which asserts an hypothesized connection between water consumption (the variable to be explained) and factors influencing the dependent variable. Parameters of demand equations in this approach are inferred via statistical inference, usually multiple regression techniques. Econometric techniques are not discussed here. Numerous textbooks present the statistical side of the issue. For the case of industrial demand, economic theory suggests that in addition to price of water, the prices of other factor inputs, type of technology or production process, product mix and output level might be significant. Data to estimate the function can come from repeated observations over time on the same activity (time series data) or simultaneous observation of many activities during the same time period (cross-sectional data). Several difficulties are typically encountered in attempting econometric estimates of industrial water demand. The major obstacle is that sufficient numbers of accurate observations for developing reliable industrial water demand functions have been difficult to obtain. To estimate a demand function requires data with a sufficient range of variation in real water prices. More generally, the number of observable cases where water is volumetrically priced is typically limited. Many industries use self-supplied ground water or surface water, and the costs for these sources may not be readily separable from general firm expenditures. However, some industries purchase water from regional water utilities on a volumetrically priced basis. Demand equation estimates based on cross-sectional data are more frequent in the literature, because time series data for a given region rarely exhibit much variation in real water rates. Because of the above obstacles, only a few studies have successfully examined industrial water demand with econometric procedures. A significant exception to the above generalization is provided by Renzetti (1992), who obtained a large data set from periodic reports required by the Canadian Government of industrial water users. As with residential use in developed country contexts, industrial water demand in Canada is reported to be somewhat price inelastic. Renzetti reports that the average price elasticity for intake water in Canadian industries was -0.38, while individual industry estimates ranged from to Renzetti also confirms that recirculation of water is a substitute for both water intake and water discharge, and concludes from this that economic incentives such as effluent fees are likely to encourage both reduced water intake and increased recirculation. Similar results were reported by Schneider and Whitlach (1991) in Columbus, Ohio. 74

93 Earlier studies include De Rooy, (1974) and Ginn, et al. (1975). Wade, et al. report that during the major California drought of , members of a sample of the California manufacturing sector spent very large amounts to avoid shortages, implying a very low absolute short run elasticity. These empirical findings of industrial water demand under conditions of metered supply contrast with anecdotal evidence of quite elastic demands in many developing countries. Where water has been unmetered or priced very low, introduction of meters or a relatively small price increase can have a major rationing effect, implying a more priceelastic industrial demand. Although quite plausible, no specific studies to document this contention have come the writer's attention. Analysts successfully performing a regression analysis which obtains a statistically reliable estimate of the marginal value or demand for industrial process water are typically content to demonstrate that the price elasticity is significantly different from zero and negative in sign. However, to derive a willingness to pay from the marginal value (demand) function, requires a further step of calculating the willingness to pay for the marginal unit for a range of potential prices. This can be done by determining the area under the demand curve via integration of the function over the desired range An Approximation Using An Assumed Demand Elasticity However, when circumstances prohibit derivation of a suitable econometric demand curve, an approximation may be acceptable. When information is limited, it can be shown (James and Lee (1971, p ) that some plausible assumptions inserted into a relatively simple equation can yield a useful approximation of incremental economic benefits. Four parameters or empirical measures are needed. The first two consist of a "price-quantity" point-an observed consumption quantity taken during a specified time period and the corresponding price in effect during that same time period. The third item is a hypothesized change in quantity that is to be valued, and the fourth is an estimate of the price elasticity of demand Finding the Area Under the Demand Function The technique described here can be applied in the special (but quite plausible) case of a constant price elasticity of demand over the relevant range of quantity change. The area under the demand curve between two specified quantities can be derived by solving a relatively simple equation reflecting the quantities at two points and the price appropriate to the initial one of those points. Assuming first, that the price elasticity of demand (e) is known and is constant over the relevant range, and is not equal to 1.0, and second, that the relevant initial price P 1 and quantity Q, can be specified, then the area under the demand curve (i.e. the desired measure of value- denoted V) for a change in quantity from Q, to Q2 is given by: V = [(Pi x QIl/E)/(l - 1/E)] [(Q t/qi - Q2/Q2u ) 75

94 The above expression represents the entire area under the demand curve for the specified quantity change, and as such, represents the value of an increment of water to the final user. This incremental value will provide a measure of benefits to compare with costs in an investment project evaluation. The parameter 1/6 (the inverse of the price elasticity of demand) is often termed by economists the "price flexibility of demand". It measures the effect of a proportional change in quantity available on the price. For purposes here, the price flexibility reflects the proportional effect of a change in water supply on the value to the user or consumer. The less price elastic the demand, the greater the price flexibility. Short run demands are usually quite inelastic-due lack of substitutes or means of adjustment-so small changes in quantity available imply large changes in value For the purpose of determining a value of industrial water for purposes of intersectoral allocation, it is then necessary to adjust the benefit estimate to reflect the derived demand for raw water in a watercourse. Comparing marginal values between sectors, such as required for assessing economic efficiency of intersectoral allocations, requires adjustments to express values in commensurate terms of form (treated or raw water), place and time. However, studies of demand for industrial water measure the willingness to pay not only for water itself, but for the services involved in capturing, transporting, treating, storing and delivering water. Therefore, in the second step, costs of these services must be deducted from estimated willingness to pay for tap water before the estimates can be properly compared with demand for instream uses or for raw water abstracted for irrigation or industry. When domestic water is priced to fully recover the costs of supplying it, the average revenue can be subtracted from the benefits estimated in step I to derive the consumer surplus imputable to raw water. In order to derive an estimate of the value of raw water (for use in studying the allocation of raw water) another step is required. This second step defines the value of raw water as the imputed consumer surplus per unit of raw water. Let the expression (Pt x Q1) represent the amount paid for the water. To obtain an estimate which represents the consumer surplus (S) associated with the change in quantity Q, to Q2, subtract the consumer's cost of water (Pi x Qi) from V (the total benefits in Eq. 4.4): S = IV- (Pi x Q)] (4.5) This measures the imputed value of a specified quantity of raw water to the user. 4.Z.2.2 Value of Raw Waterper Volumetric Unit Next, for intersectoral water allocation, it is often convenient to express the raw water value in volumetric units. Dividing the increment in consumer surplus by the assumed change in the volume of water: S/(Q2 - Q,) yields an estimate of the net benefit 76

95 per unit volume of water associated with the quantity change. This net benefit is the value of raw water, which is what is sought to compare with other raw water values Adjustmentfor Water Delivery Losses Finally, it is usually necessary to take account of water losses. The above analysis refers to the value of water to the customer derived as per the customer's meter, based on the metered sales. A final step in the approximation is needed to adjust the imputed value per cu.m. to account for losses in the delivery system. Losses due to leakages, pilferage and meter error are inevitable, but must be taken into account in an allocation formula. Multiplying the derived value per unit times the proportion available for use by customers, [or equivalently, by (I-proportion lost)], yields the final estimated value of raw water to the utility's customers Residual and Mathematical Optimization Models of Industrial Water Demand Change in Net Income The second general approach to industrial water demand employs a residual or Change in Net Income technique. Engineering design studies for the industry or plant are performed to identify a number of feasible water-using unit processes, together with alternative designs and alternative inputs and operating conditions and associated costs for each. This information is incorporated into a budget designed to ascertain the costs and returns associated with alternative technological combinations and water use amounts. The simplest form is with discrete alternatives budgeted out with partial budget formulations (Young and Gray, 1972). Spreadsheet software on desk computers have made this process relatively convenient. The residual approach (using either spreadsheet budgets or mathematical programming) to valuing water use in industry must, of course, account for opportunity costs of all nonwater inputs. However, the technique is subject to a number of pitfalls. The main difficulty follows from the fact that water, while regarded as essential to many industrial processes, typically reflects only a small portion of total input costs and incremental output. (This is in contrast to the case of crop irrigation, in which water is usually a significant input to production.) The residual is highly sensitive to assumptions made about the price to assign to nonpurchased inputs in the industrial process. The smaller the amount the marginal product of an input is, the more risk of an inaccurate estimate of the residual. For example, an industrial corporation typically is capitalized via a financial contribution from its owners as well as from borrowings from banks and bondholders. For any of a range of plausible assumptions about the opportunity cost of owned or borrowed financial capital, the estimated residual value of water might easily vary from being negative to a large positive value. 77

96 Mathematical Programming Modelsfor Estimating Industrial Water Value A related, but more sophisticated approach employs mathematical programming model formulated to select the combination of processes which maximize firm profits or minimize costs. As with programming models of agricultural water use, the definition of objective function is key. A long run formulation-with the profit function defined to measure net return after all costs- is usually appropriate for water planning exercises. The demand function is approximated by solving the model for numerous water prices and recording the amount of water required at each price. Solutions are often repeated under alternative assumptions about product prices, resource constraints and technology, to determine sensitivity to alternative model specification and to learn more about the nature of response patterns. As in agricultural sector application of mathematical models of water demand, the change in net income for an increment of water supply is imputed to the value of industrial water. Also as in irrigation studies, the objective function needs to be carefully specified so as to not attribute the productivity of other resources such as labor or investment capital to water. An improved optimization model of an industry can also be formulated so as to incorporate alternative water-use technologies, and then solved for varying constraints on water. In such event, the alternative cost approach discussed in Chapter 3, which may be used to impute a value via cost-savings between technologies, may in effect be incorporated into a mathematical programming model. The production activities which comprise the model represent a number of technical opportunities of varying cost and water use efficiency, and solution of the optimization model yields dual values which automatically impute cost savings to alternative water use levels. Clifford Russell in the 1970s developed the earliest application of mathematical programming to industrial water demands of which the author is aware. See Russell (1984) for a review of the procedure and citations of earlier references. Stone and Whittington (1984) modeled water demand for steam-powered electricity generation for a region in Poland, deriving water use predictions for a range of water costs. They employed a cost-minimization objective function, on the premise that the power plant output profile is predetermined, so that cost minimization reduces to profit maximization The "Value-Added" Approach Risks Overstating the Economic Value of Industrial Water The value-added method based on regional input-output models-criticized above in Chapter 3-has been applied in a number of instances to yield enormous but erroneous estimates of marginal industrial water values. Not only the marginal value of water but the productivity of profits, interest, depreciation, wages and salaries are assigned to water. The detailed critique from Chapter 3 (above) derived from Young and Gray (1985) is applicable to industrial sector water valuation. Table 4.3 illustrates how outlays by a representative industrial sector would be divided among purchases and value added components. 78

97 Table 4.3 Classification of Factor Inputs in a Regional Input-Output Model A. Purchased Goods and Services. Including Imports * intermediate products * raw materials * energy * transportation * Other-(transportation, spare parts, insurance, packaging) B. Value Added * salaries, wages * profits * interest * depreciation * certain taxes * net rents accruing to land, water and other natural resources From Table 4.3 it is seen that value added represents an important component of total firm purchases. Of course, the relative amounts of purchased goods and services compared to value added will vary from firm to firm and from industry to industry. However, it should be clear that a major element of upward bias is introduced if "value added" rather that the residual strictly attributable to water is taken to be a measure of marginal value productivity of water. 4.3 VALUING WATER IN HYDROPOWER GENERATION Introduction Electricity generation from hydroelectric power plants, in terms of economic value, represents on of the more important outputs of the water resource. Compared to the alternative of thermal powered generation, hydroelectric power is more flexible and environmentally friendly. Hydropower plants are longer lived, more capital intensive but require relatively less operations and maintenance expenditures. The air pollution associated with fossil-fueled plants is not a problem with hydropower (although hydropower reservoirs may slow down streams and change water temperatures). Because of the ease 79

98 with which hydro plants can increase and decrease output, hydro power is often most attractive when used for supplying power at peak demand periods of the daily fluctuations. However, because electricity demands tend to rise in winter from heating and lighting needs, hydropower demands often conflict with irrigation-and to a lesser extent residential- demands for water, both of which tend to rise in summer. Reservoir releases in winter for hydropower may preclude using that water subsequently for summer irrigation. Economic analysis can assist in making the tradeoffs between these competing uses. The economic value of water for power generation is highly site-specific. The amount of potential power generation per unit of water depends on both natural conditions at the site (the effective distance the water falls) and on the investment in water storage and generating facilities at the site and the efficiency of the generating facilities Production of Hydroelectric Power Energy production from hydropower depends on (1) the amount of water that flows through the turbines, (2) the distance that the water drops, (effective head) and (3) power plant efficiency, constrained by turbine and generator capacities. For a specific hydropower installation, energy production for a given time period (hour, day, month, year) t in kilowatt hours (KWH) is computed as: KWHt=Ht x Qt x Ef x C (4.6) subject to the following definition and constraints: H= Ept - Es (4.7) Ht 2 = Hm (4.8) KWHT < Mg (4.9) Qt <Mq (4.10) where: H = average head in feet during period t Q = flow volume (in acre feet) per unit time Ef = generator efficiency 80

99 Hm = minimum head for intake Mg = maximum generator capacity in KWV-H per unit time Mq = maximum turbine capacity in Q per unit time Ep = average pond (reservoir) elevation in feet during period Es = tailwater elevation in feet (Usually assumed constant) C = a constant to convert from acre feet into KWH (C= ). Generator efficiency is often assumed to be about 0.80 or 0 85 Equation 4.5 asserts that the production of electricity (in Kilowatt hours-kwh) is the (volume of water) times (the constant reflecting the theoretical kilowatt hours generated per unit volume per unit head) times the (generator efficiency); all times the effective distance the water falls The constraints require that estimated production be limited to maximum turbine and generator capacity The gross value of electricity output is determined by (as discussed further below) assigning an alternative cost price to units of output. Care must be taken to distinguish between base load generation and the more valuable peak load output Distinctive Issues in Hydropower Evaluation Most economic evaluation of hydropower is to assess the overall economic feasibility of a proposed investment in hydropower production capacity. Isolating the marginal value of water from the total value requires additional steps Because the electricity is produced from a combination of resources capital investment in dam, reservoir and generators plus the operating maintenance and repair costs in addition to the water, the marginal contribution of water must be derived from an additional analvtic process employing the residual technique Two steps are required to derive the economic value of water for hydropower generation. The first step is to value the electricity produced from a specific hydro plant. Because electricity is typically sold into a power grid relying on a number of sources (hydro plus thermal), it is not convenient or even possible to specifically derive the demand for the hydro portion of the region's or nation's electrical supply. Also, because electricity prices are often set by government policy, which seldom reflects the marginal cost of new supply, observed electricity rates may be inappropriate for economic evaluation. Therefore, in the first step, the value (shadow price) of electricity is usually calculated via the alternative cost technique, based on an estimate of the cost of the next likely increment of electrical power. The second step is to calculate, via the residual approach, the portion of the total value of electricity output attributable to the water used for generation Depending upon the case under study, the analyst may estimate any of several values of water fbr hydropower One pait of cases are short run and long run values. Short 81

100 run values are derived by deducting only operation, maintenance and repair (OM&R) from total value of output, and are suitable for short run reallocation decisions. Long run values are developed for long run investment and reallocation decisions, by further deducting capital investment costs (annualized equivalent costs of outlays for dam, reservoir, generating plant allocated to the power function). For a given site and electricity market, long run values are therefore less than short run value estimates. The other pair of cases refer to the value for peaking versus baseload generation. Peaking power electricity is typically more valuable than baseload generation, because of the cost of bringing less efficient and more expensive alternative thermal capacity briefly on line. Thus, water for peaking is correspondingly more valuable in peaking than in base load generation. The alternative cost valuation of peaking power is particularly difficult because of the site-specific characteristics of alternative peaking capacity and the problems of allocating fixed costs between peaking and baseload operations A Model for Deriving the Value of Water in Hydropower Generation A. C. Albery (1968) formulated a model for computing the value of water which retains its applicability. Albery's approach is explained below. (Note monetary units are shown in $, but any appropriate currency can be used). Let: G= capital cost in $ per installed kilowatt (KW)capacity of generating facilities, site work and dams; T = C = capital costs in $ per installed KW capacity (including transmission lines and substations); capital costs in $ per installed kilowatt capacity of total project (=G + T)± cac = annualized charges on capital investment, where oc is the capital recovery factor for the assumed planning period and interest rate. (cxc may be thought of as equivalent to interest and depreciation on investment),bc = e annual costs of operation and maintenance (where,b is assumed to be a constant percentage of capital cost); overall hydraulic, mechanical and electrical efficiency; f= annual capacity utilization factor (ratio of average load on the plant to installed generating capacity), h = q = effective mean head in feet (pond elevation minus tailwater elevation); flow in cfs at maximum output (all equipment assumed operating at normal full load capacity); 82

101 x = z = yf= value (in $) of one cfs of water for one year value (in $) of one acre foot of water; accounting price of electricity (in $ per KWH) at load factor f Albery's model derives the maximum willingness to pay for water given the competition of the cheapest alternative source of electricity. The value of one cfs of water for one year (denoted x) is given by: x = y4(.0848)eh - [.0848 C (Q+j3)/8760f] (4.10) where.0848 is a constant relating cubic feet of water to kilowatt hours, and 8760 is the number of hours per year, and all other symbols are as defined above. Equation expresses the value of water (in cfs for a year) as a function of head, annual capacity utilization factor, capital and operating costs and the alternative cost of electricity supplied. To further determine z (the value per unit volume -converting a flow measure to volume) divide x by , ( a constant which converts cfs per year into acre feet.) These formulas are given in English-American units of measure. The conversion to metric units is straightforward. The accounting price of electricity, determined by the alternative cost method, must be determined on a case by case basis. The OECD (1993) report "Projected Costs for Generating Electricity: Update 1992" provides procedures and current data for estimating generating costs. Dowlatabadi and Toman (1991) discuss the cost differences between alternative technologies, as well as indicating environmental costs, which in certain cases should be added to capital and operating costs. (Adding environmental damages as a cost of the alternative will, of course, properly credit hydropower with those cost savings.) The experience over the last two or three decades with the costs of electricity generation shows that predicting the alternative generating costs over a long planning period introduces an unavoidable degree of uncertainty. Technological improvement has increased the efficiency of power from thermal energy plants, a trend that can be hoped will continue (although the optimistic forecasts for nuclear power proved very wide of the mark). The real cost of energy inputs into electricity generation (coal, natural gas, etc.) has varied widely. Public policies designed to reduce air pollution can be expected to increase electricity costs, although increased decentralization and competition in the electricity industry is having the opposite effect. The analyst who undertakes to estimate the alternative cost of electricity generation "from scratch" faces a major task. However, this effort may be avoided if the forward 83

102 planning department of the relevant public or private power provider can be tapped to provide estimates of the cost of the next increment of power to the service area. 84

103 APPENDIX TO CHAPTER 4 4A.1 INTRODUCTION Several applied concepts and tools are useful to the analyst planning to estimate values of water in intermediate good-type uses. This Appendix briefly introduces several such issues, including the shadow pricing of capital and labor, and the conversion of large initial investment costs into equivalent uniform annual costs. 4A.2 SHADOW PRICING AS A COMPONENT OF WATER VALUATION Estimating the value of water, itself an exercise in shadow pricing, often requires some additional shadow pricing. This is particularly the case for intermediate good uses of water, where benefits are measured in terms of producer surpluses or net incomes. In some instances, market prices for inputs, outputs and capital may require some adjustment to reflect social values. Examples include inputs such as capital and labor and outputs such as agricultural crops or electricity. These points are addressed in detail in numerous texts on benefit-cost analysis, (for which, see, for example, Gittinger, 1983; Schmid, 1989; Pearce and Turner, 1990, Johannson, 1993) but in order to be somewhat self-contained, a brief introduction to these topics together with citations to the relevant literature warrants some attention here In the following subsections, the more general problems of shadow interest and wage rates ate discussed. 4A.3 CHOOSING THE SOCIAL RATE OF DISCOUNT In a number of frequently-encountered water valuation problems, flows of services are realized over time, and a rate of interest appears in the corresponding valuation model. In water valuation exercises, the need to specify an interest rate most often occurs where water is an intermediate good, such as in industry or agriculture, but the issue also occasionally is found where water is a consumption good. Observed market interest rates vary widely, depending on such considerations as the duration and riskiness of the debt instrument and inflationary expectations, so the choice of an interest rate for valuing water may not be immediately obvious. The economic theory of intertemporal choice has evolved to answer the question of how to aggregate single-period values over time. The standard economic theory concludes that utility maximizing individuals under competitive conditions will borrow or lend so that they equate the market rate of interest with their marginal rates of time preference between present and future consumption. Producers borrow when the anticipated rate of return on investments exceed the interest rate, while capital owners lend if the interest rate exceeds their time preference. The market interest rate would reveal preferences for the tradeoff between present versus future consumption and production opportunities, and 85

104 thus represents a "price" of capital to employ in economic evaluation. However, in the presence of taxes, inflation or other capital market imperfections, the observed market rate will be inappropriate for public policy analysis, and some adjustment is called for. The social rate of discount is the term for the shadow or accounting rate of interest used to evaluate public policies and investments which yield flows of services over time. Selection of the social discount rate is one of the most contentious and strongly debated issues in public economics, one for which the profession shows little signs of resolving. While the scope of this monograph precludes a detailed treatment, some mention of the considerations involved in the choice of social discount rate and citations to the relevant literature appropriate. (See Pearce, 1986, for a straightforward discussion of the derivation of social discount rates. Price, 1993, offers an encyclopedic but largely critical assessment-from an environmentalist perspective-of the conventional economic approach to discounting future income and cost flows.) Among the many contending viewpoints, four broad approaches to selecting a discount rate emerge. One sets the government's cost of capital (the government borrowing rate) as the appropriate rate, implicitly accepting the assumption that capital markets effectively register time preferences of producers and consumers. The cost of capital is usually taken to be the rate of interest paid by the government for long term (i.e year) debt. The U.S. government for many years has used a government borrowing rate for evaluating water resource projects. However, an immediate objection to the using the government's cost of capital as a guide is that market interest rates will, in addition to reflecting the scarcity of capital, contain an inflationary premium. Because benefit-cost analysis conventions call for real (inflation-adjusted) benefits and costs, the interest rate should also be in real terms. Howe (1971, p. 81) provides a correct formula for making the adjustment from nominal to real rates. Except for cases of high real rates of interest and high inflation, a suitable approximation is to just subtract the anticipated rate of inflation from the market rate of interest. Most authorities believe that capital markets do not function properly to reflect time preference, and merely adjusting the market interest rate for inflation is insufficient for deriving a correct" shadow interest rate. From the multitude of views on the correct discount rate, without - it is hoped - oversimplifying too much, we can say that three broad additional perspectives emerge. One of these holds that the real government borrowing rate is too high; the others contend it is too low. The social opportunity cost school emphasizes that the maximum return to scarce capital will occur when capital invested in the public sector earns the same marginal return as do investments in the private sector. However, because government taxation on private investment returns distorts private market interest rates, the social rate of return on private investments (the opportunity cost of public spending) is higher than the after-tax return influencing private markets. Because taxes are merely income transfers, the pre-tax return is the appropriate measure of social opportunity cost. This viewpoint concludes that low public discount rates preclude profitable private investments, and that market interest rates are too low to effectively measure the marginal return to capital. An after-tax rate of return is seen as a better measure of the appropriate social discount rate 86

105 An alternative approach, often called the social time preference model, rather than emphasizing the investment side of the capital market, focuses on the behavior of individual consumers as savers. Individual members of the economy are seen to be interested more in immediate rather than postponed consumption and to have short planning horizons. However, according to this perspective, society as a whole should be more forward-looking, and not display a high preference for immediate consumption. The private emphasis on early returns leads to higher interest rates than would be socially desirable. Finding a measure from which to derive a STP-based social discount rate from the capital markets is daunting; one view is that the discount rate should be based on a social value judgment reflecting society's willingness to trade off current sacrifices for long term gains. Yet another approach, one that the writer finds attractive, has been recently offered by Quirk and Terasawa (1991). These authors begin with the proposition that a government should select investments so as to maximize the social return on a fixed annual budget depending on tax receipts. They argue then, that, the opportunity cost of capital invested in a public project is not the foregone returns in the private taxpaying sector, but the return on the marginal potential public project. Finally, they suggest that this foregone return to the marginal government project is more likely to be in the region of ten percent than the lower figures reached by either the social opportunity cost or the social time preference methods. 4A.4 THE SHADOW WAGE RATE Where labor markets are not distorted by government interventions (legal prohibitions against labor mobility, minimum wages, etc.) the going market wage will reflect social opportunity costs and be appropriate for economic evaluation. However, as with the interest rate, labor costs (wage rates) also may need to be adjusted for market imperfections. Early models of imperfect labor markets emphasized that the shadow wage rate should equal the foregone marginal productivity of workers. In the presence of unemployment, the foregone marginal productivity of previously unemployed labor would be zero, and the shadow wage rate should accordingly be zero. However, an alternative perspective, and one recommended here, believes that the foregone productivity approach understates the correct shadow wage rate (Jenkins and Harberger, 1995). This model begins with the observation that, even when unemployment exists, some jobs remain unfilled, suggesting that the prevailing wage cannot entice workers into the job market. A further implication is that the shadow wage rate should perhaps approach or equal the market wage. This second model is based on the premise that work yields disutility relative to leisure, so the shadow wage should reflect the foregone marginal utility of leisure. The supply price of labor required to induce people to work on the project in question is the measure of the shadow wage (Jenkins and Harberger, 1995). The workers' preferences for location and working conditions, as well as the cost of moving to the site and living away from home should also be reflected in the shadow wage rate. Shadow wage rates should, moreover, be set to reflect differing skill 87

106 levels of the employee classes, and would also be expected to vary between different times and locations. 4A.5 CALCULATING AN EQUIVALENT UNIFORM ANNUAL COST OR BENEFIT Earlier in this report ( Chapter 2), the net present value model was illustrated as the conventional form for analyzing investment and allocation decisions. However, in many actual applications, it is more convenient to employ a variation of the net present value formula which expresses benefits and/or costs in "annual equivalents" rather than developing specific benefit and cost estimates for each year in the planning horizon. These situations are frequently encountered in practice, arising when resources for analysis are limited and/or when the forecasting of benefits and costs into the distant future is fraught with uncertainty. rule: Consider the formulation in equation 4A. 1, which is a form of the net present value Zt=1...T Bt /( + i) t > Et = I.T CI/(1 + i) (4A. 1) If the analyst judges it appropriate to assume that B, remains unchanged over the entire planning period (i.e. B 1 = B 2 =... = BT), then = 1...T Ct can also be converted into a corresponding equivalent stream of constant annual values. Costs normally vary greatly over the planning period; an initial period of large capital expenditure is followed by a longer period of smaller annual operating, maintenance and repair (OM&R) costs. There exists, for any such fluctuating stream of costs, an "equivalent uniform annual stream" which may be denoted Q, such that Z+I...T Q/(1 + i) t = I... T CtI(] + ) t (4A.2) Expression (4A.2) means that if a constant amount equal to Q were to be expended each year for T years, the net present value of that stream is the same as the met present value of the fluctuating stream Ct for T years. The capital recovery factor-a special discounting factor found in any full set of compounding and discounting tables (e.g. Gittinger, 1973) is used to convert the Net Present Value of Costs into an Equivalent Uniform Annual Cost. Called the CRF for short, the capital recovery factor facilitates the calculation of equal installments of amount Q necessary to amortize a given investment over a stated period (T years) at a specified interest rate. More technically, the CRF represents the fraction of a monetary unit (e.g. dollar) which at interest rate i for T years will equal one unit (dollar) of net present value. Put yet another way, it is equivalent to the annual payment required to retire a loan of one dollar in T years with compound interest on the unpaid balance. 88

107 For example, the CRF for finding the amount of each level payment (Q) made at the end of each of 20 years to recover the capital amount at the end of twenty years at 6% interest is Hence, the uniform annual equivalent cost of an expenditure whose present value is $100,00 over a twenty year period at six percent is $10,190. For feasibility tests using the Equivalent Uniform Annual Net Benefit measure, the appropriate selection criterion is equivalent to those employed for the Net present Value Rule or the Benefit Cost Ratio. Annual benefits should exceed annual costs, or the ratio of annual benefits should exceed 1 for project economic feasibility. Another frequent use of the CRF in water resource planning is for calculating the unit capital cost of water. Costs expressed in terms of unit capital costs are often easier for the nonspecialist to comprehend and to weigh than are aggregate cost estimates, such as the total outlay or the discounted present value of costs. For this purpose, an initial capital cost is converted into equal annual cost installments over the life of the investment. Dividing the annual capital cost by the expected average annual water delivery gives the estimated unit capital cost of water. 89

108 5. APPLICATIONS 2: VALUING WATER AS A PRIVATE CONSUMERS' GOOD 5.1 OVERVIEW This and the following chapters take up the approaches to estimating water values where water is a final consumption good, in contrast with the intermediate of producers' goods discussed in the previous chapter. These analyses address categories of economic benefits which derive from both withdrawal and nonwithdrawal uses and also from both private and public goods. The topic of this chapter is valuing water in municipal, primarily domestic uses. The discussion examines both developed and developing country issues. In the next chapter, the discussion turns to water quality and instream recreation values. 5.2 VALUING WATER IN MUNICIPAL USES IN DEVELOPED COUNTRIES A number of different types of uses fall into the municipal category. The main component is residential, or domestic which conventionally refers to all inside and outside uses by households. Inside uses are for drinking, cooking, sanitation, and so on, while outside residential uses include water for lawns and gardens, etc. Other components of the municipal group might include public uses, by government agencies, schools and other public services, including irrigation and care of public recreational facilities (Hanke and de Mare, 1984) and commercial (by nonmanufacturing business enterprises). It is often difficult to obtain data isolating other municipal customers from residential uses, so most analysts focus on residential use or the total of municipal consumption. (See Schneider and Whitlach, 1991, for an exception). Empirical domestic water demand studies typically postulate that the quantity of domestic water demanded per connection varies with the "appropriate" price of domestic water, prices of related goods, income of domestic water consumers, climate, and conservation policies. The demand function (the relationship between quantity taken and price) is represented graphically by the demand curve, or algebraically as: Qw = Qw (P., Pa, P; Y; Z) (5.1) where Qw refers to the individual's level of water use in a specified time period; P, refers to the price of water; Pa denotes the price of an alternative water source; P refers to an average price index representing all other goods and services; Y is the consumer's income, and Z is a vector representing other factors, such as climate and consumer preferences. Consumers are hypothesized to adjust water consumption behavior and to modify water using appliances in response to changes in domestic water price. Domestic water consumption thus varies inversely to price changes. 90

109 Early comprehensive reviews of municipal water demand studies were reported in Boland, et al (1984) and Herrington (1987). Examples of econometric studies of municipal water demand include Howe (1983), who restudied an earlier data set representing a nationwide sample of households with newer techniques. Martin and Thomas (1986) combine a range of price-quantity points from the U.S. with very high-priced situations in Kuwait and Australia. It is interesting to note their conclusion that these diverse data points trace out an approximately constant unitary elasticity demand curve. Schneider and Whitlach (1991) present one of the most comprehensive water demand studies available, and also provide an extensive survey of the previous literature. They analyzed a very large data set (some thirty years of individual accounts from a number of communities supplied by the City of Columbus, Ohio water system) and derived short run and long run demand functions for each of five sectors (residential, commercial, industrial, government and schools) as well as for the total of all metered demand accounts. Griffin and Chang (1991) confirmed from a sample of Texas counties that demand differs between winter and summer. Their study showed demand is somewhat more inelastic in winter (about -0.3) than in summer (about -0.4). Lyman (1992) compared peak with off-peak demands from a small Idaho city, finding a quite elastic response to peak prices, while long-run off-peak price elasticity was inelastic with respect to price. Hewitt and Hanemann (1995) reassessed a data set representing individual household observations, finding seasonal demand elasticities much greater than had previously been reported Econometric Methods for Domestic Water Demand Estimation Where the appropriate data are available, economic analysts usually apply econometric approaches to modeling domestic water user behavior (Kindler and Russell, 1984). This approach is statistical, making inferences from actual observations on quantities consumed, together with the corresponding data on prices, incomes, climatic factors and so on. An abstract demand function is formulated (such as shown above in Eq. 5.1) which asserts a hypothesized connection between water consumption (the variable to be explained) and factors influencing the dependent variable. Domestic water demand is very site-specific, varying with a range of natural and socioeconomic factors. Parameters of demand equations in this approach are inferred via multiple regression techniques. We turn next to a discussion of the individual elements in Eq. 5.1, and issues arising in estimating the demand function Type and Number of Observations Sufficient numbers of accurate observations on prices and water use for developing reliable water demand functions have been difficult to obtain. Mainly, this is because of the limited number of observable cases where water is volumetrically priced, although the proportion of water suppliers using meters has increased greatly in recent years, more so, of course, in the developed countries. If data from reasonably accurate metered deliveries is not available, average delivery per connection may be used as an approximation. How- 91

110 ever, in such a case, system leaks and other losses will introduce an unknown degree of inaccuracy. Demand equation estimates based on cross-sectional data are more frequent in the literature, because of the limited records for time series, but time series are also attempted. An important consideration with either type of data is obtaining a wide enough variation in the price variable to yield statistically reliable results. Since there is little variation in price charged to customers of a given supplier, this usually means a cross-section (simultaneous observation of behaviors during the same time period) of utilities must be sampled. Less frequently, data to estimate the function can come from repeated observations over time on the same utility (time series data), or a combination of cross-section and time series observations. The preferred data set would have observations on individual household behavior. However, due to the difficulty and expense of individual household observations, most analyses rely on aggregate water use patterns, such that each observation point represents the average water use per time period for a specific supplying entity Specifying the "Price" Variable: Marginal or Average Price? Analysts have debated for over a decade whether the "marginal" price or the "average" price is the preferable measure to represent the price in the econometric analysis. The average price is measured by the total revenue divided by the quantity of water, while the marginal price is cost of the marginal increment as provided in the rate schedule. In contrast with the situation in a competitive market, the water supply schedule faced by a consumer is established by the monopoly water supplier, which typically set rates based upon costs of service. Rate schedules include combinations of a fixed fee with a flat rate, increasing or decreasing block rate or nonmetered (fixed fee only). Estimation of residential water demand is complicated by an administered rate schedule; if a block rate structure is employed, price is endogenous, varying with the amount of water consumed. For fully knowledgeable consumers, analysts agree that the residential water demand function is correctly specified with marginal price obtained from the respective rate schedule. However, Foster and Beattie (1981) among others, propose that average revenue represents a more accurate reflection of the price actually perceived by consumers. This school contends that because water represents a small portion of expenditures, consumers do not find it in their interest to become informed on the details of the rate schedule, so marginal price is less relevant than average price. The average revenue position seems to be supported by goodness of fit statistics in econometric studies of residential water demand. At issue in the marginal price versus average revenue specification is water consumers' knowledge and their decision mechanism, and the proper econometric specification of the demand function. Only after perceived price has been correctly specified can the effect of conservation and other demand parameters be tested. McKean, Taylor and Young (1996), however, contend that for two reasons, specifying average revenue as the price measure leads to upwardly biased estimates of 92

111 both the responsiveness of water use to price and to goodness of fit measures. One reason is that (letting R denote the utility's sales revenue) the average revenue is R/Q,. Therefore, the demand function (omitting shifter variables) is actually Q, = f(r/q.), and the variable Q, appears in both sides of the expression. The variables are not independent, and hence, the estimated coefficient on average revenue and accordingly, the price elasticity of demand is biased upward. Also upward biased are the goodness of fit statistics. A second difficulty with average revenue occurs in the most frequently encountered form of rate structure: one characterized by a fixed monthly fee in addition to the block rate which varies with amount taken. When a fixed monthly fee is included in the specification of average revenue, the estimated coefficient on price and the measure of goodness of fit are further biased upward. Moreover, contend McKean, et al, (1996), ordinary least squares (OLS) estimating techniques fail to adequately capture the price schedule, and a more complex statistical procedure, such as two-stage least squares (TSLS) is preferable. (This argument does not resolve the question of whether the customer perceptions reflect average or marginal price; it does show that relying on goodness of fit measures to support the average cost position is subject to question) Additional Considerations in Specifying the Domestic Water Demand Function In addition to obtaining accurate measures of the price and quantity variables, data must be found for the important ancillary or shifter variables in the municipal demand model. These additional variables include consumer income and in data sets representing large regions, climatic factors. The possible presence of nonprice water conservation policies and incentives is an additional consideration. The main point of caution to be made here is that available data base on these variables may not coincide with the jurisdiction for which the price and quantity data are obtained, introducing a possibility of inaccuracy. Measures of household income, if available, are usually developed from periodic national censuses of political subdivisions, which may not coincide with the water supply agency's service area or the date of the water consumption data set. The degree to which an underground economy exists will also distort income estimates. Similarly, climatic instrument stations may not adequately reflect average conditions in a water service region. Analysts who attempt to control for qualitative variables such as the existence of water conservation programs face the problem of deciding what actually constitutes a conservation program Other Methods of Estimating Residential Water Demand Although econometric analyses of water use has been by far the most frequently applied approach, several other methods have been applied. These include the contingent valuation (CV) methods as well as the infrequently applied hedonic price (property market) analysis and an engineering-technical approach. Contingent valuation (CV) methods (described and evaluated in Chapter 3) have been occasionally attempted for municipal 93

112 demands in industrialized nations. Thomas and Syme (1988) carefully applied the technique to valuing residential water use in the Perth metropolitan area of Australia. If housing prices can be shown to be affected by availability of improved water supplies, the hedonic method can be applied to measure the willingness to pay for the water supply attribute. For example, North and Griffin (1993) reported a study of housing markets in a region of the Philippines (using data from 1978) to estimate the rent premium attributable to an improved water supply. A measure of imputed monthly rent (the estimated monthly rental value of homes owned by their occupants) was regressed on variables representing the attributes of the house, including size, location, number of rooms, nature of construction and type and distance of the water source from each house. A statistically significant measure of the value of access to piped water supply was obtained, although this amount was not large relative to the cost of supply. Residential water demand has been studied with an engineering-technical approach in but one case of which we are aware. Howe (1971) carefully described and priced alternative technologies for household water use, and determined the water price at which adoption of each technology would become attractive to a cost-minimizing consumer Deriving a Value of Raw Water for Domestic Purposes from a Demand Function The approaches described in the immediately previous discussion yields a demand function for domestic water at the point of delivery. However, a value for raw water is needed for evaluating proposals for intersectoral water allocation or reallocation. Two further steps are required to determine the value of an increment or decrement of raw water to retail municipal customers. The first task can be accomplished by deriving the gross benefit: achieved by integrating the inverse of the econometrically-derived demand function over the appropriate increment of quantity. In the absence of a demand curve for the specific situation, an approximation can be developed. The approximation requires knowledge of the price elasticity of demand as well as the estimated water use at a specified marginal price. For example, if - as was shown earlier for industrial demands - the analyst can assume the relevant demand function exhibits a known, constant price elasticity of demand, and a point representing water use at a given price can be identified, the formula given above (in Chapter 4, Section 4.2.1) can be solved to estimate benefits. The log-log functional form, which when fitted to appropriate data by regression techniques, yields coefficients on the variables which are the elasticities, typically yields a good and often the best statistical fit to domestic water demand data. The next step is to adjust the consumer demand function to reflect the derived demand for raw water in a water course. This step is necessary for the purpose of determining a value of domestic water for purposes of intersectoral allocation. Studies of demand for tap water must measure the willingness to pay not only for water itself, but for the services involved in capturing, transporting, treating and storing water. Therefore, in the second step, costs of these services must be deducted from estimated willingness to 94

113 pay for tap water before the estimates can be properly compared with demand for instream uses or for raw water abstracted for irrigation or industry. When domestic water is priced to fully recover the costs of supplying it, the average revenue can be subtracted from the consumers surplus estimated in step 1 to derive the net consumer surplus imputable to raw water. (See Booker and Young, 1994, for an application of this point to competing demands in the Colorado River in the western U.S.). Gibbons (1986) has reported water values which perform this conversion of tap water demand into derived demand for raw water. 5.3 MEASURING BENEFITS OF DOMESTIC WATER SUPPLY RELIABILITY In addition to the quantity of domestic water, customers value the reliability with which water is available. The reliability dimension is not captured in the conventional water demand function, which typically assumes full reliability. Because increasing reliability comes at increasing costs, estimates of the customer benefits from increased reliability are of interest. At least two cases can be envisioned for which the potential for changed reliability might have value. One is the hour to hour or day to day reliability of municipal water supplies to domestic customers. Some third world municipal water supply systems, for example, are not able to deliver water so that even their customers with piped residential connections can reliably obtain water on demand throughout the full 24- hour day, every day of the week (Nickum and Easter, 1994). Customers in such situation show a willingness to pay for water-supply reliability in several ways. One is by investing in in-home storage tanks-which are filled whenever water is available. Another is installing suction pumps to pull water from the distribution system, which reduces pressure for others, and increases risks of contamination of the main system. Where economically and hydrologically feasible, households may invest in shallow domestic wells. Altaf, Jamal, Whittington and Smith (1993) successfully measured willingness to pay for improved reliability from a sample of Pakistani households with private connections which typically supplied water only a few hours per day. A contingent valuation question of the form: "What is the most your household would be willing to pay to ensure water was available from your private connection 24 hours per day, every day of the week, with good pressure?" The responses indicated that the sample members would be willing to pay more than double the existing tariff for improved reliability. The second instance is valuing the reliability of water supplies throughout the range of climatic fluctuations-particularly, of course, during droughts. Clear evidence of this value is found in water markets in the western United States. Property rights in water are assigned according to a prior appropriation ("first in time-first in right) system, such that the highest reliability of supply pertains to those with the earliest priority of right. Market prices in those instances vary significantly with priority-the highest price, of course, being paid for earliest priority rights (Colby and Bush, 1993). Economic studies of the value of domestic water supply reliability during droughts appear to be not numerous, although interest has increased in recent years. Two which appear in the literature employ the contingent valuation method. Carson (1991) reports 95

114 on a survey of California voters regarding their annual willingness to pay for water supply reliability, finding that median annual household willingness to pay to avoid shortages was $83 for a mild shortage up to $258 for the most severe case. Howe and Smith (1994) studied the value of reliability in three small cities in Colorado, in the southwestern United States. A relatively sophisticated form of the contingent valuation method (CVM) was developed. Respondents to a mail survey were asked both willingness to pay (WTP) and willingness to accept compensation (WTA) questions. For two cities whose water supplies exhibited low reliability, a conservative estimate of aggregate WTP was insufficient to cover costs of improved reliability. For the third city, possessing at present a very high reliability, the cost savings from reduced reliability were found to be more than sufficient to cover aggregate WTA through reduced water bills. 5.4 RURAL DOMESTIC WATER DEMAND IN DEVELOPING COUNTRIES A large proportion of the earth's population - particularly in rural portions of developing nations - is without access to piped potable water supplies Improved access to potable water will yield health benefits and reduced water carrying effort - particularly for women. A traditional approach has been to assume that for low income villagers, only provision of the most basic services -hand pumps or public taps - would be warranted. Although governments and donor agencies make considerable investment in potable water supplies in rural areas of developing countries each year, the overall record of success is disappointing Village water systems are frequently inadequately maintained and often used incorrectly Many are abandoned soon afler installation. Although there are many reasons for the limited success of rural water supply programs, it has become clear that better understanding of the preferences and behavior regarding water supply of rural household will contribute to more economical and effective programs. Moreover, it is evident that traditional water demand models used in industrialized nations are inapplicable for assessing village water demands, and improved methods must be developed. During the past decade, a concerted effort has been mounted under sponsorship of international donor and lending agencies to formulate and test improved approaches. The most complete discussion of the procedures and findings of this research program is found in Whittington and Swarna (1994). (See, also Briscoe, et al., 1990; Whittington and Choe, World Bank Water Demand Team, 1993; and other references cited below). Whittington and Swarna (1994) nicely summarize the program on measuring economic benefits of potable water supply projects in developing countries. The authors discuss and illustrate the use of cost saving, contingent valuation and hedonic property value methods. There are two components of the economic benefits received by a household from installation of an improved water supply. First is the monetary value of savings in resources used to obtain water prior to the new system. (These savings may be from not having to purchase water from vendors, or from not having to fetch water from a distant source or from not having to boil water,) The second benefit component is the consumer surplus from the increased water purchased at a lower total cost. However, the above model is incomplete in the village situation because it overlooks the problem of 96

115 choice of water source. In the without-project situation, households may obtain water from several sources: among them might be open wells, public taps on a limited distribution system or water vendors. The cost, quality, distance from household and reliability may differ among these sources. Whittington and Swarna (see also Mu, Whittington and Briscoe, 1990) argue that a discrete-continuous decision process must be incorporated to successfully model village water demand. Whittington (1988) provides a practical and very useful handbook to guide the design and conduct of household surveys of willingness to pay in developing countries. The report addresses questionnaire design and implementation of the survey - including sampling, translation of the questionnaire and recruitment and training of enumerators. An appendix provides examples of questionnaires used in Haiti, Nigeria and Tanzania. (See also Casley and Lury, 1987, and Buzzard, 1990, for additional discussion of the particular problems of data collection in developing countries.) Finally, note that in addition to evaluating demand for potable water in developing countries, contingent valuation methods have also been employed to estimate willingness to pay for sanitation services. For example, Whittington, et al. (1992) studied the demand for improved sanitation services in Ghana. 97

116 6. APPLICATIONS 3: VALUATION OF SOME INSTREAM PUBLIC GOODS: INSTREAM FLOWS, WATER QUALITY IMPROVEMENT AND FLOOD RISK REDUCTION This chapter addresses measurement of public and quasi-public goods benefits relating to some issues of instream water management. The topics discussed are the application of valuation techniques to: 1) enhancement of water-based recreation and amenities; 2) water quality improvement (including recreation, waterborne diseases and waste load dilution) and to 3) flood risk reduction. These cases, differing from the private good benefits of the previous two chapters, all tend to exhibit benefits of a public or quasipublic good nature. The several different approaches to measuring benefits and costs of instream and public benefits commonly employed -- the contingent valuation, the travel cost and the hedonic pricing methods -- were discussed and evaluated in Chapter 3. These techniques are the most developed in the environmental economics literature (Freeman, 1993; Braden and Kolstad, 1991), so this Chapter only briefly surveys a representative group of applications. 6.1 DERIVING THE VALUE OF WATER IN RECREATIONAL AND AMENITY USES: OVERVIEW. An increasingly important type of economic benefits from water is its value for recreation, aesthetics, and fish and wildlife habitat. The populace of developed countries more and more often choose water bodies for recreational activities. In developing nations, as income and leisure time grow, water-based recreation is also of growing significance for their own citizens, and also often provides a basis for attracting the tourist trade. Also, significant instream values are found as habitat for wildlife and fish. Water quality is a significant component in cases associated with recreation or aesthetic enjoyment of water in its natural surroundings, so measuring benefits of water quality improvement activities is an important issue. Recreational and aesthetic values tend to be nearer the public good end of the private good-public good spectrum. Enjoyment of an attractive water body does not necessarily deny similar enjoyment to others. However, congestion at special sites, such as waterfalls, may adversely affect total enjoyment of the resource. The conventional valuation of water-based recreational experiences measures economic values in visitor day terms. This may be sufficient for evaluating water investment decisions. However, for reallocation, estimates of the marginal value of water are more appropriate. Thus, in many cases, evaluation of the marginal value of water for recreation requires a two-step procedure. This section first addresses the overall issue of measuring benefits of water-based recreation, and then turns to estimating the marginal value of changes in water supply. 98

117 In passing, it is noted that contingent valuation has been applied to a number of cases involving nonuse benefits. (Recall that nonuse values are benefits received from knowing that a good exists, even though the individual may not ever directly experience the good. Preserving an endangered species is an example). Loomis, et al. (1991) surveyed households in California to determine willingness to pay to protect and expand wetland as well as to reduce contamination of wildlife habitat, reporting that California households would pay $254 per annum in additional taxes for an increase in wetland acreage with an associated 40 percent increase in bird population. Cummings, Ganderton and McGuckin (1994) report an average willingness to pay for preservation of endangered fish species of $3.42 per household in a sample taken from a southwestern city. Loomis and Larson (1994) measured economic values of potential increases in gray whale population. Two samples of California residents were studied with CVM survey techniques, with reported values of $25. for actual whale vewers and $16. for a random sample of households for a 50% increase in population. 6.2 VALUING INSTREAM FLOWS FOR OUTDOOR RECREATION The level of flows in stream has an immediate direct effect on the utility derived from many outdoor recreation activities (Loomis, 1987; Brown, Taylor and Shelby, 1991) and a longer term impact on ecosystem status. Flow levels can directly affect quality of boating experiences, success at sport fishing, scenic beauty and swimming and wading possibilities. Longer term effects are seen on the maintenance of form and function of rivers for fish habitat. As demands for instream flows increasingly compete with offstream values, economists have increasingly sought to develop and refine techniques for estimating the marginal economic contribution of instream flows. Much like the intermediate good cases discussed above, many other inputs are required to "produce" recreational services. In this case, a derived value might be estimated by deducting the input costs from the total value of recreation. However, this is not a very satisfactory solution, and more a direct approach may be preferable. Alternatively, and preferably, the CVM technique can be designed to estimate the marginal value of instream flows. Respondents are asked to directly value increments or decrements of flow (or reservoir level). Daubert and Young (1981) and Loomis (1987) are among the examples of this approach. Photographs of varying levels of instream flows, combined with descriptions of the implications for fishing or other recreational experiences, may be used to elicit marginal values of increments or decrements of flows for such water-based activities fishing, boating, rafting or streamside recreation. Loomis, (1987a), Ward, (1987), Johnson and Adams (1988), Hansen and Hallam (1991) and Duffield, Neher and Brown (1992) are other examples of applications of CVM to valuation of instream flows. These studies tend to report that instream flow values at certain times and places may exceed those for offstream use in agriculture, suggesting that a shift in incentives might lead to improvements in net economic benefits. Cooper and Loomis (1993) however, found with a multi-site travel cost model that the imputed marginal value to wildfowl hunters of water delivered to central California wildlife refuges was relatively 99

118 small, and in only one refuge of six did the marginal value approximate the value in agriculture. Related studies of the value of enhancement of recreational fisheries are found in Harpman, Sparling and Waddle (1993) and D. M. Johnson, et al, (1995). A topic somewhat related to instream flows is the willingness to pay of home buyers for proximity to and level of lakes. Lansford and Jones (1995) studied residential house sales in an area surrounding a lake near the city of Austin in central Texas. Applying the standard hedonic price analysis approach, the value of proximity to the lake was found. More interestingly, the authors were also able to show that sales prices varied systematically with the level of the lake at the time of sale, which permitted an estimate of the marginal value of water to purchasers. 6.3 VALUATION OF WATER QUALITY IMPROVEMENT Estimating the economic benefits of water quality improvement policies present a relatively difficult task. Benefits are a measure of damages avoided from a water quality enhancement program. Damages from a degradable effluent -- the case of most interest for recreation -- depend on a number of factors such as the distance downstream, water temperature, rates of flow and the quality of the receiving waters Recreational Benefits of Water Pollution Control Early studies of the effects of water pollution in the humid eastern U. S. concluded that the costs of water quality improvement for municipal and industrial users would greatly exceed the benefits, and that the primary damages are in the form of reduced value of recreational services. Attempts to estimate water quality damages to industrial water users found limited effects. For many industrial processes, such as cooling, water quality has little impact. In other cases, such as for food processing, the quality requirements are high enough that treatment is nearly always necessary. Moreover, treatment costs are said to be not particularly sensitive to water quality (Turner, 1978). Several studies of water quality benefits have measured recreational benefits of water quality improvement. If the sites vary according to, say, water quality, it might be possible to also infer the incremental value of the improved quality from a travel cost analysis. An early effort was reported by Feenberg and Mills (1980), who formulated a logit model of site visitation in the Boston area. In a carefully documented and statistically sophisticated study, Smith and Desvouges (1986) developed estimates of the value of improved water quality on a sample of U.S. Army Corps of Engineers reservoirs. In addition to contingent valuation and simple travel cost models, Smith and Desvouges developed a generalized travel cost model designed to infer the value placed on water quality improvements by recreationists. However, the generalized travel cost model yielded results that were implausibly large, illustrating the problems of extrapolating beyond the range of the site characteristics from which the model was estimated. Mitchell and Carson (summarized by Carson, 1991) attempted to measure the value of achieving 100

119 each of several degrees of improvement in water quality (boatable, fishable and swimmable) at the national level. Estimated annual benefits in the United States of nationwide swimmable quality waters were in the $20 billion range, which the authors concluded was not enough to justify the cost of attaining that standard. Hedonic pricing has been applied in a few cases to estimate water quality benefits. D'Arge and Shogren (1988) measured the value of water quality for recreational home owners on a pair of neighboring lakes in Iowa, which exhibited sharply differing water qualities. The authors regressed observed house prices near both lakes against various attributes of the sample, including area under roof, date of construction, quality of construction, size of lot, proximity to a lake, and so on. Sales prices for recreational homes on the lake with better quality water were higher, controlling for other factors, than prices of homes on the other lake. A dummy (binary) variable for the site with higher quality water provides a measure of the willingness to pay for improved water quality. Applications of the hedonic pricing serves mainly to confirm that consumers do in fact exhibit a positive willingness to pay for environmental improvement. The approach requires unique data sets, which are unlikely to be available to use for deriving specific benefit estimates for water planning Economic Benefits of Reducing Waterborne Health Risks Several recent studies have addressed the question of measuring economic damages from outbreaks of nonfatal waterborne diseases and of chemical pollutants. The most extensive of such efforts was a study of an outbreak of waterborne giardiasis which affected several thousand people in Luzerne County, Pennsylvania in by Harrington, Krupknick and Spofford (1991). Little previous economic work had been done on such nonfatal diseases. Harrington, et al investigated two categories of benefits for study. One is the valuation of acute morbidity. They hypothesized that willingness to pay to avoid acute illness. This raises questions of valuing the direct disutility of illness, medical expenses and the value of lost time for work and leisure. A second category of benefits is reduced costs of averting behavior. These avoidance costs are the costs of actions people take to reduce their exposure to environmental contaminants. Because people will incur avoidance costs only up until the point that marginal avoidance costs equal marginal damages avoided, avoidance costs are not a full measure of damages. Harrington, et al estimated costs of boiling water and of obtaining uncontaminated water as their estimate of avoidance costs. [See also Laughland, et al (1996) for another analysis of avoidance costs in the context of preventing giardiasis; Abdalla, et al, 1993 for evaluation of chemical contamination of ground water; and Lee and Moffitt (1993) for further theoretical development of avoidance costs.] 101

120 6.4 THE ECONOMIC BENEFITS OF ALLOCATING WATER FOR WASTE DILUTION: AN ALTERNATIVE COST APPROACH Polluted waters can be improved by diluting the degraded water with a source of higher quality. Release of dilution water can yield economic benefits by either reducing damages to subsequent water users or by reduction in the costs of treating effluents. Provision of dilution water incurs corresponding costs. These costs can be the cost of constructing storage for release as dilution water. More likely, they are the foregone benefits of alternative instream uses of water for hydropower, recreation or withdrawal for irrigation, municipal or industrial purposes. Interest in the potential economic benefits of waste dilution arose in the late 1960's in the United States, as one response to growing concern over environmental degradation, a concern that led to the passage of environmental quality legislation. Several studies undertook to measure the benefits of waste load dilution to compare with costs of water storage. Because of the limited development of techniques for direct estimation of damage functions at that time, a form of the alternative cost approach was selected. Waste treatment before discharge was the alternative abatement method chosen in those early studies. The conclusions of these analyses were that dilution was a very low-valued method of pollution damage reduction. Alternatively put, waste treatment was found to be a relatively less expensive approach, when compared to constructing water storage primarily for the purpose of waste dilution. These findings probably account for the limited interest in this subject in recent years. Merritt and Mar (1969) developed the basic approach to valuing waste load dilution for a river basin case study. Gray and Young (1974) adapted the method for forecasting the value of dilution water for several major river basins in the U. S. Both of these studies focussed on diluting biochemical oxygen demand (BOD) loadings, although the model could be employed for other pollutants. Merritt and Mar defined the marginal value of dilution water as equivalent to the incremental cost of achieving the same quality of water as that which would be obtained by the release of the marginal unit of dilution water of a specified quality--that is, the marginal cost of treatment. A target standard of ambient water quality is specified. To correctly account for the diluting effect of added water, the physical configuration of the river and the location of waste discharges is incorporated into the analysis. Note that the Merritt and Mar formulation is a long-run model; it assumes that resources and time are available for investment in waste treatment plants to solve the water quality problem. The method is less applicable in a short-run situation, (one where treatment capacity could not be brought into production within the relevant time frame). A similar limitation would exist in developing country contexts, where construction of waste treatment capacity might not be possible, even in the longer run, due to financial constraints. In such situations, the value of dilution water would have to be estimated by other techniques, such as contingent valuation. 102

121 6.5 BENEFITS OF FLOOD RISK REDUCTION Governments throughout the world expend resources to change flow regimes and adopt policies to influence behavior of floodplain occupants, so as to make the best use of valuable floodplains and to reduce losses to their citizens. Public floodplain management programs produce benefits in the form of reduced (not zero) risk of flood hazards. Flood risk reduction benefits tend to be public goods, once flood control services are produced, all floodplain occupants can benefit from those services, and individuals residing or conducting business on a floodplain cannot be readily excluded from enjoying floodplain management benefits. Floodplain management benefits are measured as the difference between expected flood losses with versus without protection. The evaluation, is of course, site specific, depending on both hydrologic conditions and the nature and density of present and prospective human activity on the floodplain. The methods for estimating the economic benefits of flood risk reduction are similar to those used in other contexts, but because of the problems of the public's imperfect knowledge of flood probabilities and likely damages, and the potential for intangible impacts including the risk of death, their application presents a number of difficult and contentious aspects. Howe and Cochrane (1992) provide a conceptual discussion of the process of estimating probable damage functions for natural hazards, including floods. In comparison to that on environmental quality benefits, the literature on measuring flood risk reduction benefits is relatively limited. The principal technique for estimating urban flood risk reduction benefits has been the property damage avoided (PDA) approach, which reflects the present value of real (inflation-free) expected property damages avoided by the project or policy. The replacement and repair costs to buildings and other property and structures, with and without the flood hazard, are estimated for each of a number of river flow levels. The estimated annual benefit for a given flow level is the expected (probability-weighted) difference between repair costs with and without the flood management project of policy. Each flow is weighted by its probability of occurrence, and the benefits over all flows summed to estimate the expected benefit for each year. The annual benefits must be estimated for each year of the planning period, incorporating predicted changes in economic activity on the floodplain over time. More detailed descriptions of the approach can be found in the U.S. Water Resources Council (1983), or Penning-Rowsell, et al., (1992, Chapter 5). Shabman (1994), Howe and Cochrane (1992), and others note that the PDA method has been criticized for not incorporating nonproperty effects, such as individual and community disruption, medical expenses, productivity losses and preflood anxiety. Shabman (1994) provides probably the first comparison of alternative flood risk reduction benefit measures for the same locality. Estimates using PDA, land price analysis, and contingent valuation (CVM) methods were each developed for a portion of Roanoke, Virginia in the eastern United States. It was found that benefit estimates varied significantly across techniques. In particular, the CV approach yielded willingness to pay estimates which did not systematically increase with an increasing probability of inundation. 103

122 REFERENCES Abdalla, C. W., B. A. Roach and D. J. Epp Valuing Environmental Quality Changes Using Averting Expenditures: An Application to Groundwater Contamination. Land Economics 68 (2: ). Adamowicz, W. L., J. J. Fletcher and T. Graham-Tomasi Functional Form and the Statistical Properties of Welfare Measures. Am. J. Agr. Econ. 71 (2: ). Adams, R. M. and T. D. Crocker Materials Damages. In J. B. Braden and C.D. Kolstad, (eds). Measuring the Demand for Environmental Quality. Amsterdam, Elsevier Science Publishers. (pp ) Aheam, M. C. and U. Vasavada, eds Costs and Returns for Agricultural Commodities: Advances in Concepts and Measurement. Boulder, Co, Westview Press. Albery, A. C Forecasting Demand for Instream Uses. In W.R.D. Sewell, B.T. Bower, et al., Forecasting the Demands for Water. Department of Engineering, Mines and Resources, Ottawa, Canada Altaf, M. A., D. Whittington, H. Jamal, J. L. Liu and V. K. Smith Rethinking Rural Water Supply Policy in the Punjab Pakistan. Water Resources Research. 29(7: ). Anderson, Elizabeth, Value in Ethics and Economics. Cambridge, MA, Harvard University Press. Anderson, G. D. and R.C. Bishop The Valuation Problem. In D.W. Bromley, ed. Natural Resource Economics: Policy Problems and Contemporary Analysis. Kluwer Nijhoff Publishing, Boston, Anderson, R.L The Irrigation Water Rental Market: A Case Study. Agricultural Economics Research. 8 (2: 54-58) Anderson, R. L "A Simulation Program to Establish Optimum Crop Patterns on Irrigated Farms Based on Pre-Season Estimates of Water Supply." Amer. Jour. Agri. Econ Bash, P. K. and R. A. Young The Role of the South Platte Tributary Aquifer in Northeastem Colorado Irrigated Agriculture: Results of a Survey. Completion Report No. 173, Colorado Water Resources Research Institute, Ft. Collins. Bentkover, J.D., V.T. Covello and J. Munpower, eds., Benefit Assessment: The State of the Art. Dortrecht, Reidel Publishers. Berck, P., S. Robinson and G. Goldman The Use of Computable General Equilibrium Models to Assess Water Policies. In Dinar, A. and D. Zilberman. eds. The Economics and Management of Water and Drainage in Agriculture. Boston, Kluwer Academic Publishers. 104

123 Bergman, H. and J.-M. Boussard Guide to Economic Evaluation of Irrigation Projects (Revised Version). Paris, OECD. Bemardo. D. J., N. K. Whittlesey, K. E. Saxton and D. L. Bassett, An Irrigation Model for Managing Limited Water Supplies. Western J. Agr. Econ. 12 (1: ). Bishop, R. C. and T. A. Heberlein, Measuring Values of Extra-Market Goods: Are Indirect measures Biased? American Journal of Agric. Econ. 61 (5: ). Bishop, R. C. and T. A. Heberlein The Contingent Valuation Method. In: R. L. Johnson and G. V. Johnson, Economic Valuation of Natural Resources Issues, Theory and Applications. Boulder, CO, Westview Press. Bockstael, N. E., W. M. Hanemann and C. L. Kling Estimating the Value of Water Quality Improvements in a Recreational Demand Framework. Water Resources Research 23 (6: ). Boggess W, Lacewell R, Zilberman D (1993) Economics of Water Use in Agriculture. In: Carlson G. A., D. Zilberman and J. A. Miranowski, (eds) Agricultural and Environmental Resource Economics. Oxford University Press, New York, p 319 Boland, J. J., et al "Influence of Price and Rate Structure on Municipal and Industrial Water Use". Fort Belvoir, VA. Contract Report to U.S. Army Corps of Engineers. Booker, J. F Hydrologic and Economic Impacts of Drought under Alternative Policy Responses Water Resources Bulletin 31 (5): Booker, J. F. and R.A. Young Modeling Intrastate and Interstate Markets for Colorado River Water Resources. Jour. Environ. Econ. and Management. 26 (1: ). Booker, J. F and B. G. Colby, Competing Water Uses in the Southwest: Valuing Drought Damages. Water Resources Bulletin 31 (5): Bowen, R. L. and R.A. Young Financial and Economic Irrigation Net Benefit Functions for Egypt's Northern Delta. Water Resources Research, 21 (8: ). Bower, B. T., J. Kindler, C. S. Russell and W. R. D. Sewell Water Demand. In J. Kindler. and C. S. Russell, eds. Modeling Water Demands, Academic Press, London. Braden J. B. and C. D. Kolstad, (eds) Measuring the Demandfor Environmental Quality. Amsterdam, Elsevier Science Publishers. Bradford S. and J. Letey (1992) Simulated Effects of Water Table and Irrigation Scheduling as Factors in Cotton Production. Irrig Sci 13: 10. Briscoe, J., et al., Toward Equitable and Sustainable Rural Water Supplies: A Contingent Valuation Study in Brazil. World Bank Economic Review 4 (2: ). 105

124 Brown, M. L Farm Budgets: From Farm Income Analysis to Agricultural Project Analysis. Baltimore, Johns Hopkins University Press (for the World Bank). Brown, T. C., J. G. Taylor and B. Shelby, Assessing the Direct Effects of Instream Flow on Recreation: a Literature Review. Water Resources Bulletin 27 (6: ). Burrell, A. and B. Hill The Problem of Imputing a Charge for Farmers' Management. Journal of Agricultural Economics. May. Burt, O.R Economics of Conjunctive Use of Ground and Surface Water. Hilgardia (Berkeley, University of California Agricultural Experiment Station). Burt, O.R Allocation of Capital Costs in Enterprise Budgets. in Ahearn, M.C. and U. Vasavada, eds Costs and Returns for Agricultural Commodities: Advances in Concepts and Measurement. Boulder, Co, Westview Press, pp Burt, O.R. and D. Brewer, Estimation of Net Social Benefits from Outdoor Recreation. Econometrica 39 ( ). Buzzard, S., Surveys: Avoiding the Common Problems. in K. Finsterbusch, J. Ingersoll and L. Llewellyn, eds., Methods for Social Analysis in Developing Countries. Boulder, CC, Westview Press Cardon, G. E. and J. Letey (1992a) Soil-based Irrigation and Salinity Management Model: I. Plant Water Uptake Calculations. Soil Sci Soc Am J. 56:1881. Cardon, G. E. and J. Letey (1992b) Soil-based Irrigation and Salinity Management Model: II. Water and Solute Movement Calculations. Soil Sci Soc Am J. 56:1887. Carruthers, I., and C. Clark (1981) The Economics of Irrigation. Liverpool, Liverpool University Press Carson, R. T Constructed Markets. In J. B. Braden and C.D. Kolstad, (eds). Measuring the Demandfor Environmental Quality. Amsterdam, Elsevier Science Publishers. Carson, R.T and K.M. Martin Measuring the Benefits of Freshwater Quality Changes: Techniques and Empirical Findings In A. Dinar and D. Zilberman. eds. The Economics and Management of Water and Drainage in Agriculture. Boston, Kluver Academic Publishers. Carson, R. T, N. E. Flores, K. M. Martin and J. L. Wright Contingent Valuation and Revealed Preference Methodologies: Comparing the Estimates for Quasi- Public Goods. Land Economics 72 (1 :80-99). Cooper, J. and J. B. Loomis Testing Whether Waterfowl Hunting Benefits Increase with Greater Deliveries to Wetlands. Environmental and Resource Economics 3(4: ). 106

125 Creel, M.and J. B. Loomis Recreation Value of Water to Wetlands in the San Joaquin Valley: Linked Multinomial Logit and Count Data Trip Frequency Models. Water Resources Research 28 (10: ). Casley, D. J. and D. A. Lury (1987) Data Collection in Developing Countries, Second Edition, Oxford and New York, Oxford University Press. Chaudhry, M. A. and R. A. Young Valuing Irrigation Water in Punjab Province Pakistan: A Linear Programming Approach. Water Resources Bulletini, 25( ). Clarke, R Water: The International Crisis. Cambridge, MA MIT Press, 193pp. Clawson, M Method of Measuring the Demand for and Value of Recreation. Reprint No. 10, Resources for the Future, Washington, DC. Clawson, M. and J. L. Knetsch, The Economics of Outdoor Recreation. Baltimore, Johns Hopkins University Press. Colby, B. C. and D. B. Bush Water Rights Transactions: Market Values and Price Dispersion. Water Resources Research 29(6: ). Cropper, M. L. and W. E. Oates Environmental Economics: A Survey. Journal of Economic Literature, 30 (2: ). Crouter, J Hedonic Estimation Applied to Water Rights Markets. Land Economics, 63 (3: ). Cummings, R. G., D. S. Brookshire and W. D. Schulze Valuing Environmental Goods: A State of the Arts Assessment of the Contingent Valuation Method. Totawa, NJ, Rowman and Allanheld Publishers Cummnings, R. G., P. T. Ganderton and T. McGuckin Substitution Effects in Contingent Valuation Estimates. Am. J. Agric. Econ. 76 (2: ). d'arge, R. and J. Shogren Nonmarket Asset Prices: A Comparison of Three Valuation Approaches. In Valuation Methods and Policy-Making in Environmental Economics. H. Folmer and E. van lerland. Amsterdam, Elsevier. Daubert, J. T. and R. A. Young Recreational Demand for Maintaining Instream Flows: A Contingent Valuation Approach. Am. Jour. Agric. Econ. 63 (4: ). Debertin, D. L Agricultural Production Economics. New York, Macmillan. De Rooy, J The Price Responsiveness of Industrial Demand for Water. Water Resources Research 10 (3: ). Diamond, P.A. and J. A. Hausman Contingent Valuation: Is Some Number Better than No Number? J. Econ. Perspectives 8 (4: 45-64). Dillman, D. A Mail and Telephone Surveys. the Total Design Method. New York, Wiley. 107

126 Dinar, A. and D. Zilberman. eds. The Economics and Management of Water and Drainage in Agriculture. Boston, Kluwer Academic Publishers, Dinar A. and J. Letey (1994) Evaluation of Management and Policy Issues Related to Irrigation of Agricultural Crops. In: Tanji, K. K. and B. Yaron (eds) Advanced Series in Agricultural Sciences 22. Springer., Berlin Heidelberg. Dinwiddy, C. and F. Teal, Principles of Cost-Benefit Analysis in Developing Countries. Cambridge, UK, Cambridge University Press. Dixon, J.A., L. F. Scura, R.A. Carpenter and P B. Sherman Economic Analysis of Environmental Impacts, Second Edition. London, Earthscan Publications, Ltd. Dowlatabadi, H. and M.A. Toman Technology Optionsfor Electricity Generation: Economic and Environmental Factors. Washington, DC, Resources for the Future. Dudley, N_ J "A Single Decision-Maker Approach to Irrigation Reservoir and Farm-Management Decision-Making." Water Resources Res Duffield, J. W., C. J. Nehrer and T.C. Brown Recreation Benefits from Instream Flow: Application to Montana's Big Hole and Bitterroot Rivers. Water Resources Research 28 ( ). Duffield, J. W., T.C. Brown and S. D. Allen Economic Value of Instream Flow in Montana's Big Hole and Bitterroot Rivers. Research Paper RM-317, Rocky Mountain Forest and Range Experiment Station, U. S. Forest Service, Ft. Collins, Colorado. Easter, K.W. and R. Hearne, Water Markets and Decentralized Water Resources Management: International Problems and Opportunities. Water Resources Bulletin 31 (1):9-20. Eckstein, Water Resource Development: The Economics of Project Evaluation. Cambridge, MA. Harvard University Press. Eidman, V. R. et. al., Commodity Costs and Returns Estimation Handbook (Draft Report of American Agricultural Economics Association National Task Force on Commodity Costs and Returns Measurement Methods). Department of Agricultural and Applied Economics, University of Minnesota, St. Paul, (June). (Processed). Feenberg, D. and E. S. Mills, Measuring the Benefits of Water Pollution Abatement, New York, Academic Press. Fletcher, J.J., V.L. Adamovicz and T. Graham-Tomasi, The Travel Cost Model of Recreational Demand: Theoretical and Empirical Issues. Leisure Sciences 12 ( ). Foster, H., and B. R. Beattie "On the Specification of Price in Studies of Consumer Demand under Block Price Scheduling." Land Economics 57:

127 Fowler, F. J. Jr Survey Research Methods, Second Ed., Newbury Park, CA, Sage Publications. Franklin, D. R. and R. K. Hageman Cost Sharing with Irrigated Agriculture: Promise versus Performance. Water Resources Research 20 (8: ). Freeman, A.M., III The Measurement of Environmental and Resource Values: Theory and Methods. Washington, D.C., Resources for the Future. Freeman A.M. and W. Harrington Measuring Welfare Values of Productivity Changes. Southern Economics Journal, 56 (4: ). Gibbons, D.C The Economic Value of Water, Washington, DC, Resources for the Future. Gittinger, J.P Economic Analysis of Agricultural Projects. Baltimore, Johns Hopkins University Press (for the World Bank). Gittinger, J. P., ed., Compounding and Discounting Tables for Project Evaluation. Washington, DC, World Bank. Gleick, P. H. ed., Water in Crisis: A Guide to the World's Fresh Water Resources. New York, Oxford University Press. Gramlich, E. M., A Guide to Benefit-Cost Analysis, 2nd. Edition. Englewood Cliffs, NJ. Prentice Hall. Gray, S. L. and R.A. Young The Economic Value of Water For Waste Dilution: Forecasts to Journal of the Water Pollution Control Federation 46(7: ). Gray, S. L. and R.A. Young Valuation of Water on Wildlands. in G. L. Peterson and A. Randall, eds., Valuation of Water on Wildlands. Boulder, CO., Westview Press. Griffin. R. C and C. Chang Seasonality in Community Water Demand. Western J Agric. Econ. 16 (2): Hamilton, J. R. and N. K. Whittlesey Energy and Limited Water Resource Competition. in N. K. Whittlesey, ed., Energy and Water Management in Western Irrigated Agriculture. Boulder, CO, Westview Press, pp Hanemann, W. M Welfare Evaluations in Contingent Valuation Experiments with Discrete Responses. Am. Jrnl. Agric. Econ. 66 (2:332-41). Hanemann, W. M Willingness to Pay and Willingness to Accept: How Much Can They Differ? Am. Econ. Rev. 81 (3:635-47). Hanemann, W. M Valuing the Environment through Contingent Valuation. J1 Econ. Perspectives, 8 (4:19-43). Hanke, S. H. and L. de Mare Municipal Water Demands. In J. Kindler and C.S. Russell, eds, Modeling Water Demands, Academic Press, London,

128 Hansen, L. T. and A. Hallam National Estimates of the Recreational Value of Streamflow. Water Resources Research, 27 (2): Hedges, T.R Farm Management Decisions. Prentice Hall, Englewood cliffs NJ. Harberger, A. C Project Evaluation: Collected Papers. Chicago, IL, Markham. Hamrington, D. H Management Issues Relating to Policy Analysis. in Ahearn, M.C. and U. Vasavada, eds Costs and Returns for Agricultural Commodities: Advances in Concepts and Measurement. Boulder, Co, Westview Press, pp Hartman, L. M. and R. L. Anderson Estimating the Value of Irrigation Water from Farm Sales in Northeastern Colorado.. Farm Econ. 44 (1 : ). Hausman, J.A. ed. (1993) Contingent Valuation: A Critical Assessment. Amsterdam, North Holland Publishers. Heady, E Economics ofagricullural Production and Resource Use. Englewood Cliffs, NJ, Prentice Hall. Hedges, T.R Farm Management Decisions. Englewood Cliffs, NJ, Prentice Hall. Hanley, N. and C. L. Splash Cost-Benefit Analysis and the Environment Brookfield, VT, Elgar Publishing. Harpman, D. A., E. W. Sparling and T. J. Waddle A Methodology for Quantifying and Valuing the Effects of Flow Changes on a Fishery. Water Resources Research 29 (4: ). Harrington, W., A. J. Krupknick and W. 0. Spofford, Jr Economics and Episodic Disease: Benefits of Preventing a Giardiasis Outbreak, Washington, DC, Resources for the Future. Herfindahl, 0. C. and A. V. Kneese Economic Theory of Natural Resources. Columbus, OH, Merrill Publishing Co.406 pp. Herrington, P. R Pricing of Water Services. Organization for Economic Cooperation and Development, Paris. OECD Publications. Hewitt, J. A. and W. M. Hanemann (1995) A Discrete/Continuous approach to Residential Water Demand under Block Rate Pricing. Land Economics 71 (2: ). Hexem, R. W. and E. 0. Heady, 1978 Water Production Functions for Irrigated Agriculture. Iowa State University Press, Ames, Iowa. Hirshleifer, J., J. DeHaven and J. Milliman Water Supply University of Chicago Press. Howe, C. W The Role of Technological Change in Municipal Water Demand. in Howe, C. W. et al, The Impacts of Technological Changes, Public Policies and Changing Market Conditions on the Water Use Patterns in Selected Sectors of the U.S. Economy, U.S. National Water Commission, Report No. NWC - EES: ; by Resources for the Future, Washington, DC,. 110

129 Howe, C. W Benefit-Cost Analysis for Water System Planning. Water Resources Monograph No. 2. American Geophysical Union, Washington, DC. Howe, C. W The Impact of Price on Residential Water Demand: Some New Insights. Water Resources Research 18 (4: ). Howe, C. W. and H. C. Cochrane Guidelines for the Uniform Definition, ldentitfication and Measurement of Economic Damages from Natural Hazard Events. Special Publication 28, Institute of Behavioral Science, University of Colorado, Boulder, CO. Howe, C. W. and M. G. Smith The Value of Water Supply Reliability in Urban Water Systems. Jrnl. of Environ. Econ. andmanagement. 26(1:19-30). Howitt, R. E., W. D. Watson, and R. M. Adams "A Reevaluation of Price Elasticities for Irrigation Water." Water Resources Research Huffmnan, W.E Costs and Returns: A Perspective on Estimating Costs of Human Capital and More. in Ahearn, M.C. and U. Vasavada, eds Costs and Returns for Agricultural Commodities. Advances in Concepts and Measurement. Boulder, Co, Westview Press, pp Jenkins, G. P. and A. C. Harberger Manual: Cost-Benefit Analysis of Investment Decisions (Draft, ca. January, 1995). Cambridge, MA, Harvard Institute for International Development for World Bank Johansson, Per-Olov, The Economic Theory and Measurement of Environmental Benefits. Cambridge, UK, Cambridge University Press. Johansson, Per-Olov, Cost-Benefit Analysis of Environmental Change. Cambridge, UK, Cambridge University Press Johnson, D. M., R. J. Behnke, D. A. Harpman and R. G. Walsh Economic Benefits and Costs of Stocking Catchable Rainbow Trout: A Synthesis of Economic Analysis in Colorado. North American Jrnt. of Fisheries Management 15 : Johnson, N. S. and R. M. Adams The Benefits of Increased Streamflow. Water Resources Research 24 (11: ). Johnson, R. L., N. S. Bregenzer and B. Shelby. Contingent Valuation Question Formats: Dichotomous Choice vs. Open-Ended Responses. In: R. L. Johnson and G. V. Johnson, Economic Valuation of Natural Resources Issues, Theory and Applications. Boulder, CO, Westview Press. Just, R. E., 1991 Estimation of Production Systems with Emphasis on Water Productivity. In: A. Dinar and D. Zilberman (eds) The Economics and Management of Water and Drainage in Agriculture. Kluwer, Boston. Just, R.E., D.L. Hueth and A. Schmitz Applied Welfare Economics and Public Policy. Prentice Hall, Englewood Cliffs, NJ. Kanarck, A. M Economics and the Real World. Philadelphia, University of Pennsylvania Press. 111

130 Kindler, J. and C.S. Russell, eds Modeling Water Demands. London, Academic Press. Knapp, K. C. and D. Wichelns, Dynamic Optimization Models for Salinity and Drainage Management. in K. K. Tanji, ed., Agricultural Salinity Assessment and Management. New York, American Society of Civil Engineers. Knetsch, J Environmental Policy Implications of Disparity between Willingness to Pay and Compensation Demanded Measures of Value. Jour. Environ. Econ. and Management 18 (2: ). Kopp, R. J. and V. K. Smith, eds Valuing Natural Assets: The Economics of Natural Resource Damage Assessment. Washington, DC, Resources for the Future. Krutilla, J. V. and 0. Eckstein 1959 Multiple Purpose River Development. Baltimore, Johns Hopkins University Press. Lansford, N. J., Jr. and L. L. Jones Recreational and Aesthetic Value of Water Using Hedonic Price Analysis. J Agr. and Resource Econ. 20 (2) Lee, L. K. and L. J. Moffitt Defensive Technology and Welfare Analysis of Environmental Quality Change With Uncertain Consumer Health Impacts. Am. J Agr. Econ. 75 (2: ). Letey, J. and A. Dinar (1986) Simulated Crop-water Production Functions for Several Crops when Irrigated with Saline Waters. Hilgardia 54(1):1 Letey, J, K. Knapp, and K. Solomon (1990) Crop-water Production Functions under Saline Conditions. In: Tanji K (ed) Agricultural Salinity Assessment and Management. American Society of Civil Engineers, New York Loomis, J.B. 1987a. The Economic Value of Instream Flow: Methodology and Benefit Estimates for Optimum Flows. Jour. Environmental Management 24 (2: ). Loomis, J.B. 1987b. Balancing Public Trust Resources of Mono Lake and Los Angeles' Water Right. Water Resources Research 23 (8: ). Loomis, J.B., M. Hanemann, B. Kanninen and T. Wegge Willingness to Pay to Protect Wetland and Reduce Wildlife Contamination from Agricultural Drainage. in A. Dinar and D. Zilberman, eds. The Economics and Management of Water and Drainage in Agriculture. Boston, Kluwer Academic Publishers. Loomis, J. B. and D. M. Larson Total Economic Values of Increasing Gray Whales Populations: Results from a Contingent Valuation Survey of Visitors and Households. Marine Fisheries Economics 9: Lyman, R.A Peak and Off-Peak Residential Water Demand. Water Resources Research, 28(9): Martin, W. E., et al Saving Water in a Desert City. Resources for the Future, Inc., Washington, DC. 112

131 Martin, W. E. and J. F. Thomas Policy Relevance in Studies of Urban Water Demand. Water Resources Research 22( ) McConnell, K. E The Economics of Outdoor Recreation. In A. V. Kneese and J. L. Sweeney, eds., Handbook of Natural Resouirce and Environmental Economics, Vol. II, Amsterdam, Elsevier McConnell, K. E Indirect Methods of Natural Resource Damage Assessment. In R. J. Kopp, and V. K. Smith, eds 1993 Valuing Natural Assets: The Economics of Natural Resource Damage Assessment Washington. DC, Resources for the Future. McKean, J. R., R. G. Taylor and R. A Young Estimating Residential Water Demand in Colorado Under Alternate Price Specifications (Manuscript Submitted for Publication. Colorado State University) McKean, R Efficiency in Governmenit Through Systems Analysis. New York, Wiley. McLean, Douglas Environmental Values and Economic Trade-offs: Conflict and Compromise. In J.M. Gillroy, ed., Environmental Risk, Environmental Values anid Political Choices: Beyond Efficienicy Trade-offs in Policy Analysis. Boulder, Westview Press Merritt, L. B. and B. W. Mar Marginal Value of Dilution Water. WVater Resources Research, 5 (6: ). Michelsen, A. M. and R. A. Young Optioning Agricultural Water Rights for Urban Water Supplies During Drought. AmericanJ. Agric. Econ. 75(4: ) Michelsen, A. M., Administrative, Institutional and Structural Characteristics of a Functioning Water Market. Water Resources Bulletin 30 (6; 1-12). Miller, R. E. and P. D. Blair, Input-Ouitpuit Analysis: Foun7dationrs and Extensions. Englewood Cliffs, NJ, Prentice-Hali. Milliman, J. W Land Values as Measures of Primary Irrigation Benefits. J. Farm Econ. 41 (2: ). Mitchell, R.C. and R. T. Carson, Using Surveys to Value PuJblic Goods: 7The Conltingent Valuationi Method Baltimore, Johns Hopkins University Press for Resources for the Future. Moore, C. V., P. Sun, and J. H. Snyder "Effects of Colorado River Water Quality and Supply on Irrigated Agriculture." Water Resources Res Moore, M. R., N. G. Gollehon and D. H. Negri (1994) Alternative Forms for Production Functions of Irrigated Crops. JAgric Econ Research 44(3): 16 Morgan, M.G and M. Henrion Uincertainty: A Guide to Dealing with Uncertainty in Quantitative Risk and Policy Analysis Cambridge, Cambridge University Press. Morgenstern, The Accuracy of Economic Observations, Second Edition Princeton, Princeton University Press. 113

132 Mu. X., D Whittington and J. Briscoe, 1990 Modeling Village Water Demand Behavior: A Discrete Choice Approach Water Resouirces Research 26 (4: ) Munasinghe, M., Water Supply and Environmental Management. Developing World Applications, Westview Press. Boulder, CO Nickum, J. E. and K. W. Easter AMetropolitan Wf'aler UIse Conflicts inl Asia anld the Pacific, boulder Colorado. Westview Press. Nieswiadomy, NI.L Estimating Ulrban Residential Water Demand: Effects of Price Structure. Conservation and Education. Wl'aler Resource.s Journal, 28 ( ) North, J. H. and C. Griffin, 1993 Water Source as a Housing Characteristic: Hedonic Property Valuation and Willingrness to Pay for Water WVater Resources Research 29( ) Organization for Economic Cooperation and Development (OECD) Management of Water lprojects. )ecision-making ansd Im',estmewi Appraisal. Paris. OECD Publications Organisation for Economic Cooperation and Development (OECD) Projected Costs for Generatin7g Electricity Update 1992 Paris, OECD Publications. Palm-quist, R. B 1991 Hedonic Methods. In Braden J.B and C.D Kolstad, (eds). Measuring the l)emand for tnviwronmental Qualiry Amsterdam, Elsevier Science Publish ers Pearce, D. W (ost-benefit Analysis, Second Edition (Revised,). London, Macmillan Pearce, D W and A. Markandya Environmental P'olicy Benefits. Monetary Taluationi Paris, Organization for Economic Cooperation and Development. 89 pp. Pearce, D W. and R. K. Turner Economnic.s (?f n7atural resources an7d the enivironlmen7t Johns Hopkins University Press, Baltimore, MD Pearce, D W Fconomwc V7alues, an7d the Natural World. Cambridge, MA, MIT Press Pearce, D W and J J Warford 1993 World Without A`nd: IE7conomics, Environment al7d Sustainahle l)evelopvmen Oxford UIniversity Press (for the World Bank), New York Penning-Rowsell., E. C., C. H. Green, et al., The Economics of Coastal Management: A Manual of lbenefit Assessment Techniqutes London, Bellhaven. Price, C lime, Discounting and Value Oxford. UK, Blackwell Publishers. Quirk, J. and K. Terasawa, Choosing a Government Discount Rate: An Alternative View. J Lty. flcon. and A4,anagem ent 20 (I 16-28) I 14

133 Randall, A Valuation in a Policv Context. in D.W. Bromley, ed. Naturlal Resource Economics: Policy Problemxs CSand Coniemporary A nalysis. Kluwer Ni jhoff Publishing, Boston, Randall, Alan Resource Econoomics 2nd ediiuon,). New York, John Wiley Rae, A N 1994 Agricuiltural Management 4nalysis:.Acrli'ly AnalY.is andci l)ecixsom- Making. Wallingford. Oxon., U5K. CAB International Rea, L M. and R. A. Parker, 1992 Designig and Conducting Suirvev Research, San Francisco, Jossey-Bass Renn, Concepts of Risk A Classification fn S Krimsky and D Golding, eds SXocial Iheorie.s of Ri.sk Praeger Publishiers. Westport, CT, pp Renzetti, S Estimating the Structure of Industrial Water Demands The Case of Canadian Manufacturing. Land P&conomics 68(4): Rhoads, S. S The Economist's l'iew of the World: Governments, Markets Canld Public Policy. Cambridge. UK, Cambridge University Press. Rogers, P Anmerica'y Water: hfederal Roles atncd Re.sponsihihiiies Cambridge, MA. MIT Press, 285 pp Rosegrant, M. W and H. P Binswanger 1995 Markets in Tradeable Water Rights: Potential for Efficiency Gains in Developing Countries W'orld l)c1'elopme11. 22( ). Russell C.S 1984 Programming Models for Regional Water Demand Analysis. In J Kindler and C S Russell, eds, Mof4de/ing W%aler Demands, Academic Press, London, 1984 Salant. P and D A Dillman, Ho, 1o Conduct youir Own,Survey New York. Wiley. Saliba, Bonnie Colby and D.B. Bush 1987U Waier Markets in Theorv anid Pr-actice:. Market Translers anl1d Public Policv Boulder. CO. Westview Press. Scheierling, S M (1995) Measuring Demand for Irri-ation Water and Foregone Direct Benefits of Irrigation Water Transfers Ph D Dissertation. Colorado State University, Fon Collins Schmid, A.A Benefit-Cost Analysis: 4 Political l conlomy Aspproach Boulder, CO, Westview Press. Schneider, M L. and E E. Whitlach User-specific Water Demand Elasticities JrnIz. WYater Resources Platnitng andcl Maanagement. 17 ( 1: 52-73). Schultze, W.D Use of Direct Methods for Valuing Natural Resource Damages In R. J. Kopp, and V K. Smith, eds azluinlg Natural.4ssets: The AEconoincs q/ Natural Resource D)amage Assessmnen. Washington. DC, Resources for the Future 115

134 Shabman. L Measuring the Benefits of Flood Risk Reduction. In: Y. Y. Haimes, D.A. Moser and E. Z. Stakhiv, eds. Risk-Based Decision Making in Water Resources VI, New York, American Society of Civil Engineers. pp Smith, V.K A Conceptual Overview of the Foundations of Benefit-Cost Analysis. in J.D. Bentkover, V.T. Covello and J. Munpower, eds., Benefit Assessment: The State of the Art. Dortrecht, Reidel Publishers Smith, V.K Can We Measure the Economic Value of Environmental Amenities? Southern Economic.Journal, (April) Smith, V.K Nonmarket Valuation of Environmental Resources: An Interpretive Appraisal. Land Economics 69(l): 1-26 Smith, V.K. and W H. Desvouges, Measuring Water Quality Benefits. Kluwer Nijhoff Academic Publishers, Boston. Smith, V K and Y Kaoru 1990 Signals or Noise? Explaining the Variation in Recreational Benefit Estimates Am. 1J Agr. Econl. 71 ( ). Solley, W.B., R.R. Pierce and H. A. Perlman 1993 Estimated Use of Water in the United States, 1990 Circular 1081, U.S Geological Survey, Washington, DC. Spulber, Nicolas and Asghar Sabbaghi Lconomics of Water Resources: From Regulation to Privatization. Kluwer Academic Publishers, Norwell, MA. Steiner, P.O The Role of Alternative Cost in Project Design and Selection. Quarterly Jour. of Fcon. 79 (3) August. Stewart, D.W Secondary Research: Information So02rces and AMethods. Sage Publications, Beverly Hills, CA. Stone, J. C. and D. Whittington, Industrial Water Demands. in J. Kindler and C.S. Russell, eds, Modeling Water Demands, Academic Press, London, Taylor, R.. G. and R. A. Young, 1995 Rural-to-Urban Water Transfers: Measuring Foregone Direct Benefits of Irrigation Water Under Uncertain Water Supplies. J. Agr. and Resource Econ. 20 (2) Thomas, J.F. and G-J, Syme Estimating Residential Price Elasticity of Demand for Water A Contingent Valuation Approach. Water Resources Research 24 (11: ). Torell, A., J Libbin and M. Miller The Market Value of Water in the Ogallala Aquifer. Land Economics 66 (2: ). Turner, R. K Water Pollution. In D. W. Pearce, ed. The Valuation of Social Cost. London. George Allen and Unwin. U.S. Water Resources Council Economic and Environmental Principles and Giuidelines for Water and Related Land Resource Implementation Studies. Govt. Printing Office, Washington, DC 116

135 van der Tak, H. G The Economic Choice Between Hydroelectric Power and Thermal Power Developments. Washington, DC, World Bank Staff Occasional Papers No 1. Varian, H. R., Intermediate Microeconomics: A Modern Approach, 3rd ed New York, W_ W Norton. Vaux, H. J., Jr, and W 0. Pruitt (1983) Crop-Water Production Functions. Advances in Irrigation 2:61 Wade, W. W, J. A. Hewitt and M. T. Nussbaum Cost of Industrial Water Shortages. Report to California Urban water Agencies by Spectrum Economics. San Francisco. Walsh, R.G Recreation Economic L)ecisions: Comparing Benefits and Costs. Venture Publishing, State College, PA 638 pp. Walsh, R. G., D. M. Johnson and J. R. McKean Benefit Transfer of Outdoor Recreation Demand Studies: Water Resouirces Research 28 (5: ). Ward, F. and J. B. Loomis The Travel Cost Demand Model as an Environmental Policy Assessment Tool: a Review of Literature. Western J Agric. Econ. 11(2) Ward, F Economics of Water Allocation to Instream Uses: Evidence from a New Mexico Wild River. Water Resources Research 23 (3: ). Ward, W. A. and B. J. Deren The Economics of Project Analysis. A Practitioner's Guide Washington, D.C. World Bank Whittington, D "Guidelines for Conducting Willingness to Pay Studies for Improved Water Services in Developing Countries" Arlington. VA, Water and Sanitation for Health (WASH) Project. Prepared for U.S. Agency for Internationai Development. (Processed). Whittington, D et al. (1992) Household Demands for Improved Sanitation Services in Kumasi, Ghana: A Contingent Valuation Study. Water Resources Research 29 (6: ). Whittington, D. and K. Choe Economic Benefits Available from the Provision of Improved Potable Water Supplies. Technical Report No Water and Sanitation for Health (WASH) Project. Arlington, VA, Prepared for U.S. Agency for International Development. Whittington, D. and V. Swarna The Economic Benefits of Potable Water Supply Projects to Households in Developing Countries. Economic Staff Paper No. 35, Asian Development Bank, Manila. Willet, G. S Measurement Issues Relating to Short-run Farm Financial Decisions. in Ahearn, M.C. and U. Vasavada, eds 1992 Costs and Returns for Agricultural Commodities: Advances in C-oncepts and Measurement. Boulder, Co, Westview Press, pp

136 Williams, H. P Model Building in Mathematical Programming, 3rd ed. New York, Wiley. Winpenny, J. T The Economic Appraisal of Environmental Projects and Policies: A Practical Guide. Paris, Organization for Economic Cooperation and Development. World Bank Water Resources Management: A World Bank Policy Paper. Washington, DC. (140 pp.) World Bank Water Demand Research Team The Demand for Water in Rural Areas: Determinants and Policy Implications. World Bank Research Observer 8 ( ). Xu, F., R. C. Mittlehammer and P W. Barkley Measuring the Contributions of Site Characteristics to the Value of Agricultural Land. Land Economics 69 (4: ). Young, H.P., N Odaka and T. Hashimoto Cost Allocation in Water Resources Development Water Resources Research 18(3): Young, H.P ed., 1985 Cost Allocation: Methods, Principles and Practices. Amsterdam, Elsevier Science Publishers. Young, R.A 1978 Economic Analysis and Federal Irrigation Policy: A Reappraisal. WesternJ. Agric. Ecotn. 3 ( ) Young, R.A Managing Aquifer Overexploitation: Economics and Policies, In Ian Simmers, F. Villaroya and L F. Rebollo. eds. Selected Papers on Aquifer Overexploitation, Heise Publishers. H annover, FRG. Young, R.A Water Economics. In L. W. Mays, ed., Handbook of Water Resources. New York, McGraw Hill. Young, R. A., and J.D Bredehoeft "Digital Computer Simulation for Solving Problems of Conjunctive Groundwater and Surface Water Systems." Water Resources Res Young, R.A. and S. L. Gray (with R B Held ana R S. Mack) Economic Value of WFater: Concepts and Empirical Estimates Report to the U.S. National Water Commission. (Nationai Technical Information Service Publication PB , Sptingfield. VA Young, R A and S L Gray 1985 Input-Output Models, Economic Surplus and the Evaluation of State or Regional Water Plans. Water Resources Research, 21 (12: 1819-i823) Young, R.A. and R. Haveman Economics of Water Resources: A Survey. In A. V. Kneese and J. L. Sweeney, eds.. Handbook of Natural Resource and Enivironmental Econiomics. Vol. II, Amsterdam, Elsevier. 118

137 Distributors of CHMA & Ecnorc PRpasounouSA Lwcsa Convosana Sansom SPA Na nrarooor Center W orld Pubshix Houa 35o Stoauram SU Va idat)cs1aana, 11 Book Departnment W orld Bank S.~~~~ Da Fo S, ng A 0682 Athens CaselbaPDsrlahsSS2 PO Eo 6125 Eeerstrhd SINGAPORE. TAINAN, Van Daeta Edbti 7Te60q MYANAR, BRUNEI Ch dlacez4 Asahgate PubksNngAss CH1807 BWray Publications Eejmg Tel (1) Feaize N-0620slo6 Pacrt;ile ta LT: (021) Prices and credil JermIs vary Tel (1) a () 3A-254 F (05) 645 Fax (22) GoIr K l9 dng3 Fax. (021) F (1) Fx 5)6127Pl 2)6 91Gle t,fx fromt counitry to counitry. HOGKN,MCOS,ngapa TANZANIA Consult your local distributor COLOMBIA Asa 200 Ltd JAMAICA PAKISTAN Tel (65) Od UneradyPress before placing ant order. Inoensace Lls. Sales & Cnatmbon Dogaprkent tan asdre PoIAnheea Ltd Mra Book Agency Fax (65) aksba Suel ApsaiadoAerno Seabird House. md Old Hne Road 65, Shdhraih-e-Cad-e-Azamn ea-sa ashgaleasieanconnentclnx PO Bo Boge D E 22-2 Wyndham Street Central Kngsronl6 PO Box No 729 Dat es S9aam ALBAIBA Tel. (1) Hono Kong Tel La[ore SLOVAK REPUBLIC Tel t51) Adnn Ltd Fax (1) 2N Tel Fax Tel (42) S6-ail GTG Lld Fn (51) Perdl Rndexe Sn Fan Fan (42) Krupnska 4 Pah. 9. Sh 1 AP 4 COTE DTVOIRE URLi hftpflwjesalesoanolowonom hk JAPAN PO Box IS2 THAILAND Teana Centre d'ednton at de Ddlusron Eastem Book Sence Oxford Unnersaty Press Bratislana S Cenbta Books D>sirbubon Tel (42) 27419, AlnGnas (CEDAt HUNGARY Hongo 3-Choei. 5 Bangaiore Inwn Tel (7) S9rn Road FPu (42) P 541 Foundabon Ir Medial Borkyo -k 113 Shame Fareat Faa. ( Bangkok Abeans04 Plateau Economy Tokyo PO Boa 3033 Tr (2( ARGENTINA Tel 22524S6510 Dombo,en U Tet (03) Kaiach SOUTH AFRICA, BOTSWANA Fa, (2) ChciaxdelLitrointenaaonal Fex H-1117tBudaples Fan (03) Tel (21) Fo,sailetres An Coidoka 1877 lel at URL hnpja tnbekkoame oruvp-svl-ebs Fax (21) Odord Unruers,ty Press TRINIDAD & TOBAGO JAM Buenos Ares CYPRUS Emall oupof kf orm com pk Soulthern Atica Systernaics Studies Unr Tel (1) Center olappled Research Fax KENYA PO Boa 1tr1 '9 Wans Sreet FPa (1) CypnusCodlege AincaBoDkSerNe (EA)(Ltd PERU CapeTowsn000 Caiep9 6. Dogenes Stree. Engoan INDIA uatasnhouse, Mlangaro Streel Edtonal DesarndloSA Tor (21) TInmdad. West Indes AUSTRALJA, FII, PAPUA NEW GUINEA, PO Box 20D6 Allied Pubkshets Lid PO Box Apanado 3624 Fax (21) Te SOLOMON ELANDS, VANUATU, AND Nosea 751 Mount Road Naissl Urma 1 Fax WESTERN SAMOA Tel Madras D tel. (2) Tel (14) For subsnptlon orders D A. intomiaudn Sornces Fao Tel (44) Fao (2) Fax (14) Inter,atisat Subscnrpeon Se.o e UGANDA 646 Whrnehorse Road Faxu (44) PO Bon Gusos Ltd Mtcham 3132 CZECH REPUBLIC KOREA, REPUBLIC OF PHILIPPINES Carghall Madhsa, Bu.ikng Vctlna Natural intoaration Center INDONESIA Daepon Tradmg Co Ltd Ilmemational Booksousie Center Itm Juhatesburg 2024 PO Boa 9997 let (61) prndelis, Konskiska 5 Ft Indram[triled PO Box 34 SIne 720 Cry0and O Ten (li) Ftst 164 Jn 9 a Rd Fa, (6S) CS - i3 57 Pidau0 1 Jabsan otodur 20 YIoenda Condosminm Tower 2 Fox () Kaipala URFL:AdIpiPeeddrctcoxrnau Tel ( PO Boax 81 Seoul HVdeia Cosla comet T1eTax (41) Fax (2) Jakant Tel. (2) t4 Valeur Sr SPAIN AUSTRIA URL hlitpjwensisce Tel (21) Fn (2) Make. Metno Mamsia Mundr-Piensa Lrbrs, S A UNITED KINGDOM GIerod ard Co Fax (21)42t-1289 Ts- (2) Caslello 37 M,snto Lid Graxen 31 DENMARK MALAYSIA Fax (2) Madnd PO Box 3 A 1011 Wen Sanlunds neratur [RAN L-erursy of Malaya Coopenarrue Tel (1) AJns Han,pshir GU34 2PG Tel (l) Rorsenoems Alle 1i Ko.kab Pblrsheror BoOkshOp. Lrnlted POLAND F,, (iu Englandf faxr ()l DK 1970 Fiedenksbei 9 2 PO Boa t PO B.r 1127 Inrernalronal Pub -shrnq Serre hip lewws tsal esrmpensa Te (1420) Tel (31) Tehrrr Jalrn Patruaiu Bo. U ekna 31/37 Fei 11420) BANGLADESH Fa:r ( Tel (21) ,j Koala Lumpur Wnvzana MuncdiPensa Burouo,a Msi Inrdusirnes Deseloprment Fox 90121) Tel ( W50 Tel (2) _ Co,sell de Cent, 391 ZAMBIA Ass,starce SoLeety (MIDAS) EGYPT, ARAB REPUBUC OF -a, Fax (2 b Bwucelona U,rensry Sookshop House 5 Road 16 A Ah-ram KetaL Saen Co Puosrshn Iel (3) 4-830Di9 irear Eas4 Road CamF-s DOarr,Drea P-Aea djgasaa Sueet Kiralecd EdElambmrnlb Ae MEXICO PORTUGAL Fao ( PO Bo esaaa 1209 Cawo 6A Sireed INFOTEC r orana FPuuga -&saka Te (2) Tel ( hu hnh Delasrooa Nc U Apanado Pustal 72 B60 oua Oo Cramo 7074 SRI LANKA, THE MAt DIVES Tel (l( Ext 44d Far (2) Fa, ( Iehra, Talpan 12Ou,t-b L.ke house Sookshor Ter oo87i6l04 M-eco F To ( j PO Box 244 7BABWE BELGIUM ThelMddle EastObserer Fao Tel (5)606D001 l (U34170?o4 100 SrChnasnpalnamA Lon[ mn 7rmbabue (PI (lid jeio De Lanno y at Shenl Sueet l r k1tab saranodo trlle,it Fax (5) Gwdlr-,er Mawatha lcrle R,Od Asxrr3nn A. d. Pf 2J2 ; uuo ROMANIA C-erSnti 2 F1O Bo. ST12; 113, B,ussrLs i'1 2) IRELAND NETHERLANDS :-n-poe, De Lbnarrl Ba-esl 3 A 7-1 -rl H J 5me-n- Ter 2) ron 2) Guoemment SupplesuAge"c y De LrsdecunrOrVnI-Pub.lkates Sir Lrpsca no 25 sector 3 [ron cajole Fax (2( A,7 on lsciathall P-D Box 202 B.- lrz-l I FINLAND 4 -shalrcno Road 7480 AE aaksbeigen le (0r 613 SWEOEN Fa. (4,l S,670 BRAZIL Akartsmmen K.lnaka,ppa Drorrn 2 Tel 153) Fx 0i 31240J Frlzen Customer Senne P,ibicaco5esoTln,s inten73rra o t Oc, 23 I ) d4613)11; Fax 15357?-929S RegeY.nosoalt 12 ida FIN00371 HelenA Fan RUSSIAN FEDERATIUlNI 05 4?Sirrorr Rua PFeotGornmde 20 ler -u NEW ZEALAND rdalol-otrcvesma- To, (0-60 e Cv So Paulo. SP FPs 0) 12t14441 ISRAEL E0SCO N7 Ltd 9a Kiolpachiy Pereuiok F"x 8) Ta. il URL hnf boonet ourner IVarO Yurmor Lnraluoe Ltd Poinate Malr Bag Jk.on ha (11) PC ES ries Market Tel r95j Vi,nergren Wrllans AB FRANCE her A Auckland I, :K9! P C B,s 130 CANADA -1dc Baenk PubObarons Tel (3) Tel (9) Slne Re.-td P.ksxeg Co [s 66. a,enue dilna Fax '3) Fax ( SAUDI ARABIA OATAH ri 1294 Algoma Road Plas.r Rnk SILhoe oax (SI T Oare Onta.o0 K16 Tlel ( Ynternnabonal NIGERIA PC- B, c3196 l! Fea (1) P Box U enrty Press Umned i yadh SWITZERLAND rt T-5430 T'eloA,,,61t3 T(,reeCro-ns Suold,lg Teoc.c 11 T Lurb,ar Payot GERMANY Te! (3! Prrnale Mail Bag 5095 P-i (1) Ser,e Instluknnel 0UN-VeLag Fax (3) Ibadan c5teseae-montben. 30 Pxppolsseir Ives 55 Tel (22( Lausanne Bon Palesa,ox Au.rhontyrladdle East Fas (22) Tel (021( Tel (228) Index lstos lcon Senvies Fax (02U Fax i POO t9502jerusasenr

138

139 RECENT WORLD BANK TECHNICAL PAPERS (continued) No. 299 Tamale, Jones, and Pswarayi-Riddihough, Technologies Related to Participatory Forestry in Tropical and Subtropical Countries No. 300 Oram and de Haan, Technologies for Rainfed Agriculture in Mediterranean Climates: A Review of World Bank Experiences No. 301 Mohan, editor, Bibliography of Publications: Technical Department, Africa Region, July 1987 to April 1995 No. 302 Baldry, Calamari, and Yameogo, Environmental Impact Assessment of Settlement and Development in the Upper LUeraba Basin No. 303 Heneveld and Craig, Schools Count: World Bank Project Designs and the Quality of Primary Education in Sub-Saharan Africa No. 304 Foley, Photovoltaic Applications in Rural Areas of the Developing World No. 305 Johnson, Education and Training of Accountants in Sub-Saharan Anglophone Africa No. 306 Muir and Saba, Improving State Enterprise Performance: The Role of Internal and External Incentives No. 307 Narayan, Toward Participatory Research No. 308 Adarmson and others, Energy Use, Air Pollution, and Environmental Policy in Krakow: Can Economic Incentives Really Help? No. 309 The World Bank/FOA/UNIDO/Industry Fertilizer Working Group, World and Regional Supply and Demand Balances for Nitrogen, Phosphate, and Potash, 1993/ /2000 No. 310 Elder and Cooley, editors, Sustainable Settlement and Development of the Onchocerciasis Control Programme Area: Proceedings of a Ministerial Meeting No. 311 Webster, Riopelle and Chidzero, World Bank Lendingfor Small Enterprises No. 312 Benoit, Project Finance at the World Bank: An Overview of Policies and Instruments No. 313 Kapur, Airport Infrastructure: The Emerging Role of the Private Sector No. 314 Valdes and Schaeffer in collaboration with Ramos, Surveillance of Agricultural Price and Trade Policies: A Handbookfor Ecuador No. 316 Schware and Kimnberley, Information Technology and National Trade Facilitation: Making the Most of Global Trade No. 317 Schware and Kimberley, Information Technology and National Trade Facilitation: Guide to Best Practice No. 318 Taylor, Boukambou, Dahniya, Ouayogode, Ayling, Abdi Noor, and Toure, Strengthening National Agricultural Research Systems in the Humid and Sub-humid Zones of West and Central Africa: A Frameworkfor Action No. 320 Srivastava, Lambert and Vietmeyer, Medicinal Plants: An Expanding Role in Development No. 321 Srivastava, Srnith, and Fomo, Biodiversity and Agriculture: Implicationsfor Conservation and Development No. 322 Peters, The Ecology and Management of Non-Timber Forest Resources No. 323 Pannier, editor, Corporate Governance of Public Enterprises in Transitional Economies No. 324 Cabraal, Cosgrove-Davies, and Schaeffer, Best Practicesfor Phlotovoltaic Household Electrification Programs No. 325 Bacon, Besant-Jones, and Heidarian, Estimating Construction Costs and Schedules: Experience with Power Generation Projects in Developing Countries No. 326 Colletta, Balachander, and Liang, The Condition of Young Children in Sub-Saharan Africa: The Convergence of Health, Nutrition, and Early Education No. 327 Valdes and Schaeffer in collaboration with Martin, Surveillance of Agricultural Price and Trade Policies: A Handbookfor Paraguay No. 328 De Geyndt, Social Development and Absolute Poverty in Asia and Latin America No. 329 Mohan, editor, Bibliography of Publications: Technical Department, Africa Region, July 1987 to April 1996 No. 332 Pohl, Djankov, and Anderson, Restructuring Large Industrial Firns in Central and Eastern Europe: An Empirical Analysis No. 333 Jha, Ranson, and Bobadilla, Measuring the Burden of Disease and the Cost-Effectiveness of Health Interventions: A Case Study in Guinea No. 334 Mosse, and Sontheimer. Performance Monitoring Indicators Handbook No. 336 Francis, with Akinwumi, Ngwu, Nkom, Odihi, Olomajeye, Okunmadewa and Shehu, State Community, and Local Development in Nigeria

140 I ~~~~~~ il1 I I * ^ * I. I ~~~~~~~~~~ ' '. gi g _= s~~~~~~~~~ S. I. & F~~~~~~~I - IS _~~~~~~~~~~~~~~~~~~~~~~~702 i, , B~~~~~~~~~~~~~~~~~~~~~~~~~~ *. t